Gut Microbiome in Serious Mental Illnesses: a Systematic Review and Critical Evaluation

UC San Diego UC San Diego Previously Published Works

Title Gut microbiome in serious mental illnesses: A systematic review and critical evaluation.

Permalink https://escholarship.org/uc/item/2nx0z5sd

Authors Nguyen, Tanya T Hathaway, Hugh Kosciolek, Tomasz et al.

Publication Date 2021-08-01

DOI 10.1016/j.schres.2019.08.026

Peer reviewed

eScholarship.org Powered by the California Digital Library University of California SCHRES-08457; No of Pages 17 Schizophrenia Research xxx (xxxx) xxx

Contents lists available at ScienceDirect

Schizophrenia Research

journal homepage: www.elsevier.com/locate/schres

Gut microbiome in serious mental illnesses: A systematic review and critical evaluation

Tanya T. Nguyen a,i,⁎, Hugh Hathaway b,c, Tomasz Kosciolek c,d, Rob Knight c,e,f,g, Dilip V. Jeste a,g,h,i a Department of Psychiatry, University of California San Diego, CA, United States of America b Medical Sciences Division, University of Oxford, Oxford, United Kingdom c Department of Pediatrics, University of California San Diego, CA, United States of America d Małopolska Centre of Biotechnology, Jagiellonian University, Kraków, Poland e Department of Computer Science and Engineering, University of California San Diego, CA, United States of America f Department of Bioengineering, University of California San Diego, CA, United States of America g Center for Microbiome Innovation, University of California San Diego, CA, United States of America h Department of Neurosciences, University of California San Diego, CA, United States of America i Sam and Rose Stein Institute for Research on Aging, University of California San Diego, CA, United States of America article info abstract

Article history: Schizophrenia and bipolar disorder (BD) are associated with debilitating psychiatric and cognitive dysfunction, Received 20 June 2019 worse health outcomes, and shorter life expectancies. The pathophysiological understanding of and therapeutic Received in revised form 19 August 2019 resources for these neuropsychiatric disorders are still limited. Humans harbor over 1000 unique bacterial spe- Accepted 22 August 2019 cies in our gut, which have been linked to both physical and mental/cognitive health. The gut microbiome is a Available online xxxx novel and promising avenue to understand the attributes of psychiatric diseases and, potentially, to modify

Keywords: them. Building upon our previous work, this systematic review evaluates the most recent evidence of the gut Schizophrenia microbiome in clinical populations with serious mental illness (SMI). Sixteen articles that met our selection Bipolar disorder criteria were reviewed, including cross-sectional cohort studies and longitudinal treatment trials. All studies re- First episode psychosis ported alterations in the gut microbiome of patients with SMI compared to non-psychiatric comparison subjects Bacteria (NCs), and beta-diversity was consistently reported to be different between schizophrenia and NCs. Microbes Ruminococcaceae and Faecalibacterium were relatively decreased in BD, and abundance of Ruminococcaceae Probiotics was reported across several investigations of SMI to be associated with better clinical characteristics. Lactic acid bacteria were relatively more abundant in SMI and associated with worse clinical outcomes. There was very limited evidence for the efﬁcacy of probiotic or prebiotic interventions in SMI. As microbiome research in psychiatry is still nascent, the extant literature has several limitations. We critically evaluate the current data, including experimental approaches. There is a need for more uniﬁed methodological standards in order to arrive at robust biological understanding of microbial contributions to SMI. © 2019 Elsevier B.V. All rights reserved.

1. Introduction with SMI are prone to diseases associated with aging (Czepielewski et al., 2013; Hennekens et al., 2005; Soreca et al., 2008) and have Schizophrenia and bipolar disorder (BD) are severe neuropsychiatric twice the risk of dying from cardiovascular and gastrointestinal diseases disorders, which combined have a lifetime prevalence of 3.5% (Perälä compared to the general population (Saha et al., 2007). Physiological et al., 2007). Patients with these serious mental illnesses (SMI) face changes seen throughout the body with normal aging occur at earlier not only debilitating psychiatric and cognitive impairment but also ages, including chronic inﬂammation and oxidative stress, which has worse physical health outcomes and considerably shorter life expectan- led to the theoretical framework of SMI as disorders of accelerated bio- cies (Brown, 1997; Casey et al., 2009; Viron and Stern, 2010). SMI con- logical aging (Jeste et al., 2011; Kirkpatrick et al., 2008; Nguyen et al., tribute substantially to the global burden of disease (Whiteford et al., 2018a; Palmer et al., 2018; Rizzo et al., 2014). Given that lifespans are 2013) and rank among the leading causes of disability (Chong et al., generally increasing for the general population (Christensen et al., 2016) and mortality worldwide (Walker et al., 2015). Younger adults 2009), while the mortality gap for schizophrenia is growing (Lee et al., 2018), understanding the mechanisms of potential accelerated aging in SMI is imperative. ⁎ Corresponding author at: University of California San Diego, 9500 Gilman Drive #0664, La Jolla, CA 92093, United States of America. Despite decades of research, our understanding of the pathophysiol- E-mail address: [email protected] (T.T. Nguyen). ogy of these disorders is still limited. Genomic research has identiﬁed

Please cite this article as: T.T. Nguyen, H. Hathaway, T. Kosciolek, et al., Gut microbiome in serious mental illnesses: A systematic review and critical evaluation, Schizophrenia Research, https://doi.org/10.1016/j.schres.2019.08.026 2 T.T. Nguyen et al. / Schizophrenia Research xxx (xxxx) xxx susceptibility genes, but the results have not yet led to new therapies. seemingly discrepant findings across investigations. Although more The human gut microbiome is a dynamic population of microbes in useful insights may have been drawn from a meta-analysis of data our large intestine that form a symbiotic superorganism, with which across studies, considerable heterogeneity in study designs and meth- we have co-evolved (Dinan et al., 2014). Containing 1013 microorgan- odologies to quantify and analyze the gut microbiome and few publicly isms with over 1000 unique bacterial species containing 2 to 20 million available data from individual reports made this endeavor impractical. unique genes, the gut microbiome is a complex genomic structure with Instead, we highlighted commonalities and differences among these 100 times more genes than the human genome (Gill et al., 2006; Human investigations. Microbiome Project Consortium, 2012; Qin et al., 2010; Turnbaugh et al., 2007). Unlike the human genome, which is fixed and unchangeable, the 2. Methods gut microbiome is highly dynamic and malleable. It can be shaped by various developmental and environmental influences, such as age, ge- 2.1. Search strategy ography, cultural traditions and lifestyles (e.g., diet, cohabitation, travel), and medications (Caporaso et al., 2011; Koenig et al., 2011; We searched PubMed, PsycINFO, and Embase for articles pub- Yatsunenko et al., 2012). In fact, the overall heritability of the lished before March 7, 2019 using the following search string: microbiome is low, and the microbiomes of genetically unrelated, co- microbiome AND (schizophrenia OR psychosis OR bipolar OR serious habiting individuals is more similar than of those who are members of mental illness). Reference lists of the retrieved articles and relevant the same family but living apart (Caussy et al., 2019; Cekanaviciute review articles were cross-referenced. We examined the titles and et al., 2017; Yatsunenko et al., 2012). Furthermore, the gut metagenome abstracts of all citations and selected empirical reports based on is a better predictor of many human physical phenotypes (e.g., body our inclusion/exclusion criteria. mass index [BMI], waist circumference, glucose and high-density lipoprotein [HDL] levels, lactose consumption) than the human genome 2.2. Inclusion/exclusion criteria (Rothschild et al., 2018). Both of these characteristics of the microbiome have important implications for the development of new therapeutic Studies were selected if they met the following criteria: 1) were em- approaches. pirical studies of individuals clinically diagnosed with schizophrenia, The gut microbiome is critical in maintaining human physiology. It schizoaffective disorder, BD, or related psychotic disorders, 2) utilized regulates many metabolic processes essential for optimal health that high-throughput sequencing methods to characterize microorganisms cannot be maintained by human cells, stimulates normal immune mat- in the gut or distal large intestine, and 3) were published in English. uration, defends against pathogens, and stabilizes the gut barrier Both cross-sectional and longitudinal studies were included. We ex- (Carroll et al., 2009). SMI are characterized by increased gut permeabil- cluded review papers, meta-analyses, abstracts, case reports, and stud- ity (Severance et al., 2014, 2013, 2012). With a compromised intestinal ies exclusively using animal models. lumen, enteric microbes are exposed to systemic circulation. Gut dysbiosis may underlie the pro-inflammatory milieu and other physio- 2.3. Review process logical abnormalities that have been implicated in SMI (Hsiao et al., 2013). Moreover, the microbiome plays a major role in the development Our database search yielded 184 articles, once duplicates were re- and functioning of the central nervous system (Clarke et al., 2013; Cryan moved. The titles and abstracts of all articles were screened and, of and Dinan, 2012; Diaz Heijtz et al., 2011; Neufeld et al., 2011). Recent these, 37 were assessed for eligibility. In total, 16 met all of the above- preclinical investigations indicate that gut microbes can influence mentioned criteria for review. The PRISMA flow chart depicting infor- brain and behavior (Crumeyrolle-Arias et al., 2014; Hsiao et al., 2013; mation through different stages of the systematic review is shown in Jørgensen et al., 2015; Sampson et al., 2016; Sudo et al., 2004; Zheng Fig. 1. et al., 2016), leading to a resurgent interest in the role of gut microbes Only three articles (Evans et al., 2017; Flowers et al., 2017; Schwarz in neuropsychiatric disorders and the potential ability to improve psy- et al., 2018) overlapped with our original review (Nguyen et al., 2018b), chiatric and cognitive well-being through their manipulation. which included five papers on the microbiome; the other two investiga- Given the bidirectional communication between the gut and brain, tions were of the oropharyngeal microbiome and did not qualify for the via the “gut-brain axis,” the concept of “psychobiotics” has emerged in current review. recent years (Dinan et al., 2013; Sarkar et al., 2016; Wall et al., 2014; Zhou and Foster, 2015). Probiotics are live microorganisms that confer 3. Results abeneficial health effect. Prebiotics are nondigestible food components that are selectively fermented by intestinal microflora, which are associ- 3.1. Characteristics of reviewed studies ated with health and wellbeing (Gibson et al., 2004). Much of psychobiotic research is based on animal models, which have demon- Detailed sample and methodology characteristics for each study are strated improvements in cognition, mood, and neurophysiology follow- provided in Table 1. A summary of relevant data from the 16 reviewed ing probiotic and prebiotic treatment (Sarkar et al., 2016; Savignac et al., studies is presented in Table 2. Five studies included individuals with 2013). Human clinical investigations have begun only recently. schizophrenia and/or schizoaffective disorder, seven studies included This article updates our prior systematic review of studies of the persons with BD, one study comprised a mixed sample of patients microbiome in schizophrenia and BD (Nguyen et al., 2018b). In the with schizophrenia and BD, two included patients with first episode short time since, a number of additional empirical studies have been psychosis (FEP), and one article was of high-risk individuals. Most stud- published, and more conceptual reviews and commentaries have been ies (62.5%) sampled outpatients, while three investigations recruited in- written on the gut-brain axis and its role in mental illnesses. However, patients. A majority of studies were cross-sectional (75%). Four studies aside from our own paper, we did not find other systematic reviews of involved longitudinal assessment of the gut microbiome. All reviewed the composition of the gut microbiome in SMI and its relationship to studies had at least one comparison group: 11 compared SMI to a clinical, physical, and disease-related aspects of these disorders. This re- non-psychiatric comparison (NC) group; five involved another psychi- view is distinctly different from our previous article in that it is only fo- atric comparison group; four compared subjects pre-post treatment. Fi- cused on the gut microbiome (i.e., does not include studies of nally, studies were conducted worldwide, with 25% from the US, 44% microbiomes of other tissues/organs) and includes 13 new studies. from Asia, and 31% from Europe. We provide a narrative synthesis of what can be a complex and seem- Below we summarize findings for different clinical populations ingly contradictory field of knowledge, and we posit reasons for (Tables 3–5). Within each, we highlight cross-sectional and longitudinal

273 Records identified by systematic database searches 103 PubMed 140 Embase 30 PsycINFO Identification

184 Unique records screened after duplicates removed

Screening 89 Records excluded based on title or abstract

37 Full-text articles assessed for eligibility

21 Articles excluded with reasons:

Eligibility 2 Case report 4 Not of gut microbiome 5 Animal study 10 Abstract

16 Studies included for qualitative synthesis in the review Included

Fig. 1. PRISMA ﬂow diagram for selection of published articles for review.

findings related to 1) global community diversity, 2) taxonomic differ- For each of these methods, various levels of analysis and statisti- ences, 3) clinical characteristics associated with microbial biomarkers, cal treatment can be used to extract meaningful results. Overall pat- and 4) functional potential (if applicable). terns in microbiome variation and community structure are typically There are multiple levels on which the microbiome can be analyzed assessed by alpha-diversity and beta-diversity (Fig. 2). Alpha- and characterized. We provide a brief summary of these techniques and diversity quantifies feature diversity within individual samples, measures to provide the reader context to understand and interpret the which can be compared across groups. Various indices of alpha- findings presented below (see Knight et al., 2018 for a more compre- diversity characterize the number and distribution of species in a hensive review of best practices for analyzing microbiomes). Marker community, representing species richness and evenness. It is com- gene amplification and sequencing (e.g., 16S rRNA amplicon sequenc- monly observed that low alpha-diversity is a hallmark of dysbiosis ing) uses primers that target a specific region of a gene of interest to de- (Yatsunenko et al., 2012). Beta-diversity captures dissimilarity be- termine microbial phylogenies of a sample. These methods are well- tween a pair of samples, generating a distance matrix based on ei- tested, fast, and cost-effective for obtaining a low-resolution view of mi- ther presence-absence of quantitative species abundance data. crobial communities (often limited to a genus taxonomic level). Shot- Another approach is to examine differentially abundant taxa or func- gun metagenomics is a method of sequencing all microbial genomes tional elements (e.g., genes and pathways) between groups. How- within a sample that yields more detailed genomic information than ever, this approach to understanding differences between marker gene sequencing alone (Quince et al., 2017). It may be more ex- diagnostic groups versus controls can be challenging given that pensive and is presently less streamlined than 16S sequencing, but cap- microbiome datasets are high-dimensional and compositional. The tures all DNA present in a sample and allows for greater taxonomic compositionality problem has been discussed in our previous review resolution to species or strain level. (Nguyen et al., 2018b), and is further mentioned in Section 4.

laect hsatcea:TT gyn .Htaa,T ocoe,e l,Gtmcoim nsrosmna lnse:Asseai eiwand review systematic A illnesses: mental serious in microbiome Gut al., et Kosciolek, T. Hathaway, H. Research, Schizophrenia Nguyen, evaluation, T.T. critical as: article this cite Please Table 1 Sample and methodology characteristics of reviewed studies.

Publication Country Sample size Mean age Gender Sample characteristics Assessments Sequencing Data processing/analysis Diversity assessments

High-risk and first episode psychosis He et al., China HR: 81 HR: 21.67 HR: 41M/40F UHR: met one of following on SIPS: DSM-IV; prodromal sx (SIPS, SOPS); 16S rRNA (V4 region, 515F/806R); QIIME2 pipeline; sPLS-DA to α: observed OTUs, 2018 UHR: 19 (SD = 5.75) UHR: 15M/4F BIPS, APSS, GRDS; Outpatients; no global functioning (GAF) Sequencing; Illumina MiSeq 250 bp cluster samples; PICRUSt with Shannon Index NC: 69 UHR: 20.47 NC: 37M/32F AP, AD, anticonvulsants; no info on paired-end Greengenes and KEGG Databases β: PCoA, PLS-DA (SD = 4.57) BMI (3 KEGG levels); LEfSe NC: 23.13 HR: one first-degree relative with SZ (SD = 3.89) NC: No info on matching; No info on BMI Schwarz Finland FEP: 28 FEP: 25.9 FEP: 16M:12F FEP: 14 SZ, 4 SZP, 1 SZA, 2 BD I, 1 UD, DSM-IV; positive and negative sx qPCR for 16S primers; analysis of 7 CLC Genomics Workbench V6 α: did not assess et al., NC: 16 (SD = 5.5) NC: 8M:8F 6 PD NOS; AOS = 25.9 (SD = 5.5); (BPRS, SANS); global functioning bacterial groups [Lachnospiraceae pipeline; Sequences filtered to β: did not assess 2018 NC: 27.1 outpatients; AP: 10 olanzapine, 7 (GAF); medical hx; diet (Health (Eubacterium rectale group), ≥60 bp, sequences matching (SD = 6.0) risperidone, 8 quetiapine; BMI = Behavior and Health among the Ruminococcaceae (Clostridium human genome build 37 23.8 (SD = 4.3) Finnish Adult Population survey); leptum group), Bacteroides spp., removed, remaining reads NC: Matched by age, sex and region physical activity (Gothenburg scale) Atopobium group in addition to matched to Refseq Bacterial https://doi.org/10.1016/j.schres.2019.08.026 of residence; BMI = 23.9 (SD = 3.1) Bifidobacteria and Database (length fraction = 0.8, Lactobacillus-group (comprising of similarity = 0.8); LEfSe the genera Lactobacillus, ..Nue ta./ShzprnaRsac x xx)xxx (xxxx) xxx Research Schizophrenia / al. et Nguyen T.T. Leuconostoc, Pediococcus, and Weissella)]; NanoDrop 2000c Metagenomic; Illumina Hi Seq 2000 100 bp paired-end Yuan et al., China FESZ: 41 FESZ: 23.1 FESZ: 23M/18F FESZ: AOS = 17.9 (SD = 3.2); DOI = DSM-IV; positive and negative sx qPCR for 16S primers; analysis of 5 SPSS v24.0 statistical analysis; α: did not assess 2018 (pre-post) (SD = 8.0) NC: 20M/21F 5.9 months (SD = 3.0 months); (PANSS); medical and psychiatric bacterial taxa [Bifidobacterium spp., qPCR quantification according to β: did not assess NC: 41 NC: 24.7 inpatients; AP naïve at baseline; BMI hx; physical exam; metabolic Escherichia coli, Clostridium cycle threshold; (SD = 6.7) = 20.54 (SD = 2.52) parameters (glucose, triglycerides, coccoides group, Lactobacillus spp., NC: No info on matching; BMI = HDL, LDL, insulin, hs-CRP, SOD, Bacteroides spp.] 20.75 (SD = 2.85) HOMA-IR)

Schizophrenia Nagamine Japan SZ: 16 (pre-post) SZ: 63.0 SZ: 5M:11F SZ: no info on AOS or DOI; inpatients, Psychotic sx (BPRS); weight; Terminal restriction fragment Genemapper genotyping α: did not assess et al., (SD = 10.9) length of hospital stay = 3053 days metabolic parameters (glucose, length polymorphism analysis for software; division into 29 OTUs, β: did not assess 2018 (SD = 2805); CPZE = 729.7; BMI = triglycerides, total cholesterol, 16S primers (no info on region, no information on OTU picking 20.9 (SD = 3.7) (pre), 22.3 (SD = albumin); ADRs (fever, abdominal 516F/1492R) process 4.3) (post) pain, constipation, diarrhoea) Nguyen USA SZ: 25 SZ: 52.9 SZ: 14M:11F SZ: SZ or SZA; AOS = 21.5, DOI = DSM-IV; positive and negative sx 16S rRNA (V4 region, 501F/806RB) QIIME2 pipeline; sOTU definition α: observed OTUs, et al., NC: 25 (SD = 11.2) NC: 15M:10F 32.4; outpatients; WHO DDD = (SAPS, SANS), depression sx Sequencing: Illumina HiSeq 2000 with deblur; rarified to 7905 Shannon Index, 2019 NC: 54.7 2.01; BMI = 31.8 (SD = 5.4) (PHQ-9), psychiatric and medical 150 bp paired-end sequences per sample; Faith's PD (SD = 10.7) NC: Matched by age and sex; BMI = hx; physical and mental well-being β: unweighted 28.9 (SD = 4.0) (SF-36), medical comorbidity UniFrac, (CIRS); CHD and CVD risk Bray-Curtis (Framingham) dissimilarity Okubo Japan SZ: 29 (pre-post) SZ: 45 SZ: 11M:17F SZ: no info on AOS or DOI; DSM-V; anxiety and depression sx 16S rRNA (V3–4 region, QIIME pipeline; Sequences α: Shannon Index et al., Tx responders: (median) Tx responders: outpatients; HADS ≥ 10; CPZE = 600 (HADS, PANSS); diet Tru357F/Tru806R); filtered to ≥150 bp; Open β: UniFrac 2019 12 (IQR = 16) 3M:8F (median) (IQR = 400); BMI = 25.7 (semi-structured food frequency strain specific 16S PCR for reference OTU picking with 97% distances Tx Tx Tx (median) (IQR = 5.4) questionnaire); positive, negative, bifidobacterium breve A1 (A1F/A1R) threshold; non-responders: responders: non-responders: Tx responders: CPZE = 600 affective, and disorganized sx Sequencing: Illumina MiSeq 17 46 (median) 8M:9F (median) (IQR = 508); BMI = 26.5 (BPRS); inflammatory markers (34 (IQR = 12) (median) (IQR = 6.4) pro- and anti-inflammatory Tx non- Tx non-responders: CPZE = 643 cytokines, related molecules, responders: (median); BMI = 23.6 (SD = 5.1) e.g., ligands and receptors) 41 (median) (IQR = 16) Shen et al., China SZ: 64 SZ: 42 SZ: 36M:28F SZ: no info on AOS or DOI; ICD-10; psychiatric sx (PANSS); 16S rRNA (V3–4 region, 341F/805R) QIIME pipeline; removal of low α: number of 2018 NC: 53 (SD = 11) NC: 35M:18F outpatients; no info on medications; psychiatric and medical hx Sequencing: Illumina HiSeq 2500 quality sequences; open reads, Faith's PD, NC: 39 BMI = 23.49 (SD = 3.8) reference OTU picking with 97% observed OTUs, (SD = 14) NC: No info on matching; BMI = threshold; Rarefaction to 10,000 Shannon Index, 23.14 (SD = 2.8) sequences per sample; LEfSe Simpson, ACE, Chao1 β: unweighted laect hsatcea:TT gyn .Htaa,T ocoe,e l,Gtmcoim nsrosmna lnse:Asseai eiwand review systematic A illnesses: mental serious in microbiome Gut al., et Kosciolek, T. Hathaway, H. Research, Schizophrenia Nguyen, evaluation, T.T. critical as: article this cite Please UniFrac, PCoA Zheng China SZ: 63 SZ: 43.49 SZ: 42M:21F SZ: no info on AOS or DOI; AP: 15 DSM-IV; positive and negative sx 16S rRNA (V3–4 region, 338F/806R) QIIME pipeline; removal of low α: ACE, Chao, et al., UD: 58 (SD = 1.68) UD: 22M:36F clozapine, 14 risperidone, 9 (PANSS); depression sx (HAM-D); Sequencing: Illumina MiSeq 250 bp quality and chimeric sequences; Shannon Index 2019 NC cohort 1: 69 UD: 40.6 NC cohort 1: olanzapine, 5 chlorpromazine, 3 psychiatric and medical hx paired-end open reference OTU picking with Beta: PLS-DA NC cohort 2: 63 (SD = 11.7) 36M:33F aripiprazole, 3 quetiapine, 9 2+ AP, 5 97% threshold; Taxonomy NC cohort 1: NC cohort 2: unmedicated; BMI = 22.90 (SD = assignment with SILVA database; 39.99 23M:40F 0.32) LEfSe (SD = 1.62) UD: 19 taking medications (no NC cohort 2: further info); BMI = 22.0 (SD = 2.4) 41.8 NC cohort 1: Matched with SZ; BMI (SD = 12.3) = 23.16 (SD = 0.33) NC cohort 2: Matched with UD; BMI = 22.6 (SD = 2.5)

Bipolar disorder Aizawa Japan BD: 39 BD: 40.3 BD: 17M:22F BD: 13 BD I, 26 BD II; mood state: 23 DSM-IV; depression sx (HAM-D); RT-qPCR for 16S or 23S primers; ANCOVA adjusted for age and α: did not assess et al., NC: 58 (SD = 9.2) NC: 22M:36F depressed, 2 manic, 12 euthymic, 1 mania sx (YMRS); psychiatric and analysis of only Biﬁdobacterium and sex and BMI to compare β: did not assess 2018 NC: 43.1 mixed; AOS = 28.2 (SD = 9.4); no medical hx Lactobacillus subgroups; Yakult Biﬁdobacterium and Lactobacillus (SD = 12.9) info on DOI; outpatient; mean Intestinal Flora-SCAN abundance; partial correlation medication doses (n): CPZE = 182.9 analysis against continuous https://doi.org/10.1016/j.schres.2019.08.026 (13); imipramine equivalence = 204 variables (12); lithium = 418.8 (16);

valproate = 725.0 (8); lamotrigine xxx (xxxx) xxx Research Schizophrenia / al. et Nguyen T.T. = 186.5 (13); carbamazepine = 325.0 (4); no meds (3); BMI = 23.9 (SD = 4.7) Comparison: Matched by age and sex; BMI = 22.4 (SD = 3.8) Bengesser Austria BD depressed: 13 BD (both BD (both BD depressed (HAM-D N 10); BD DSM-IV; psychiatric and medical 16S rRNA (V1-V2 region); Ion Sequence processing: DeconSeq, α: Simpson Index et al., BD euthymic: 19 sub-cohorts): sub-cohorts): euthymic (HAM-D b 10); subtype hx; HAM-D; YMRS; BDI Torrent One Touch 2.0 Kit Acacia tool, Usearch algorithm; and Evenness 2019 41.67 25M:7F not indicated; outpatients; no info analysis: QIIME 1.8; Rarified to Index (SD = 17.51) on AOS or DOI; no info on 8000 sequences per sample β: did not assess medications; BMI = 27.99 (SD = 6.45) Coello Denmark BD: 113 BD: 31 (no SD BD: 43M:70F BD: 44 BD I, 65 BD II; newly ICD-10; depression sx (HAM-D); 16S rRNA (V3-V4 region); USEARCH 10.0, mothur 1.38 and α: observed OTUs, et al., UR: 39 provided; UR: 18M:21F diagnosed; mood states: 68 mania sx (YMRS); psychiatric and Sequencing: Illumina MiSeq (MiSeq inhouse script pipeline; removal Shannon Index 2019 NC: 77 range 26–39) NC: 30M:47F euthymic, 26 depressed, 4 manic, 8 medical hx; physical activity (IPAQ) Reagent Kit V3, 2 × 300 bp of low quality and chimeric β: weighted and UR: 28 (no SD hypomanic, 6 mixed; AOS = 17; DOI paired-end) sequences; closed reference OTU unweighted provided; = 11; BMI = 24.8 (no SD provided; picking with 97% threshold; UniFrac distances range 22–34) range 22.2–27.8) Taxonomy assignment with NC: 29 (no SD UR: unaffected first-degree relatives; SINTAX provided; BMI = 24.4 (no SD provided; range range 21.8–26.4) 24.5–40.5) NC: age- and sex-matched; BMI = 24.2 (no SD provided; range 22.0–26.3) Evans USA BD: 115 BD: 50.2 BD: 32M:83F BD: 76 BD I, 29 BD II, 10 BD NOS; DSM-IV, DIGS; depression sx 16S rRNA (V4 region) mothur 1.36.1 pipeline; α: did not assess et al., NC: 64 (SD = 12.8) NC: 24M:40F outpatients; most on more than one (PHQ-9); mania sx (ASRM); anxiety Sequencing: Illumina MiSeq sequence alignment to SILVA β: Yue and 2017 NC: 48.6 psychiatric medication; no info on sx (GAD-7); psychiatric and database; OTU picking with 97% Clayton distance (SD = 16.6) AOS or DOI; BMI = 29.3 (SD = 7.2) medical hx; physical and mental threshold; NC: No info on matching; BMI = well-being (SF-12); sleep (PSQI) 26.0 (SD = 4.6) Flowers USA BD on AP BD on AP BD on AP BD I, BD II, BD NOS (nsnot DSM-IV, DIGS; psychiatric and 16S rRNA (V4 region) mothur 1.36.0 pipeline; removal α: Simpson et al., treatment: 46 treatment: treatment: provided); outpatients medical hx Sequencing: Illumina MiSeq V2 of low quality and chimeric Diversity Index 2017 BD off AP 46.0 12M:34F AP group: inclusion defined by use of sequences; OTU picking with β: Yue and treatment: 69 (SD = 12.0) BD off AP an atypical AP (clozapine, 97% threshold; LEfSe Clayton distance BD off AP treatment: olanzapine, risperidone, quetiapine, treatment: 21M:48F asenipine, ziprasodone, lurasidone, 51.7 aripiprazole, paliperidone, and (SD = 13.5) iloperidone). Other medications: 26 AD, 32 MS, 20 lithium, 13 BZ; groups did not differ in treatment with MS or AD, though the AP group used

(continued on next page) 5 6

Publication Country Sample size Mean age Gender Sample characteristics Assessments Sequencing Data processing/analysis Diversity assessments

more BZ; no info on AOS or DOI; BMI =31(SD=7) Non-AP group: matched for previous hospitalizations and other metabolic comorbidities; BMI = 27.5 (SD = 6) Flowers USA SZ/BD on AP: 21 SZ/BD on AP: SZ/BD on AP: BD I, BD II, BD NOS, BD with DSM-IV; psychiatric and medical 16S rRNA (V4 region); mothur 1.36.0 pipeline; removal α: Inverse et al., (19 pre-post) 54 (SD = 10) 12M:9F psychosis, SZA, or SZA; outpatients; hx; physical and mental well-being Sequencing: Illumina MiSeq 250 bp of low quality and chimeric Simpson Index 2019 SZ/BD not on AP SZ/BD not on SZ/BD not on AP inclusion was defined by the use of (SF-36); 24 h dietary recall paired-end sequences; OTU picking with β: Bray-Curtis (on MS): 16 AP (on MS): (on MS): 9M:7F an atypical AP (clozapine, (ASA24); anthropometric 97% threshold; LEfSe dissimilarity 50 (SD = 15) olanzapine, risperidone, quetiapine, measurements (height, weight, or ziprasidone) or lithium and/or blood pressure) lamotrigine for at least 6 months; no info on AOS or DOI; BD subjects may overlap with Flowers et al. (2017) AP group: 9 BD I, 3 BD II, 4 SZ, 5 SZA; https://doi.org/10.1016/j.schres.2019.08.026 other medications: 9 AD, 5 BZ, 12 for hypertension, 2 for diabetes, 10 for hyperlipidemia; BMI = 30.6 ..Nue ta./ShzprnaRsac x xx)xxx (xxxx) xxx Research Schizophrenia / al. et Nguyen T.T. Non-AP group: 10 BD I, 6 BD II; other medications: 12 AD, 4 BZ, 4 for hypertension, 2 for hyperlipidemia; BMI = 31.1 Painold Austria BD: 32 BD: 41.31 BD: 18M:14F BD: BD I; inpatients hospitalized for DSM-IV; depression sx (BDI; 16S rRNA (V1-V2 region); QIIME pipeline; Sequences α: observed OTUs, et al., NC: 10 (SD = 14.73) NC: 4M:6F depressive episode; all on HAM-D); psychiatric and medical Sequencing: Ion Torrent One Touch filtered to ≥100 bp; removal of Chao1, Shannon 2019 NC: 31.4 medications: 24 atypical AP, 8 hx; metabolic parameters 2.0 Kit low quality and chimeric Index, Simpson (SD = 7.61) lithium, 11 anticonvulsants, 23 (triglycerides, HDL cholesterol, sequences; OTU picking with Index antidepressants; DOI = 17.5; BMI = fasting plasma glucose); 97% threshold β: weighted and 28.44 (SD = 6.08) anthropometric measurements unweighted NC: No info on matching; BMI = (waist-to-hip ratio, waist-to-height UniFrac distances 24.26 (SD = 3.76) ratio) Vinberg Denmark MZT Affected: 71 MZT Affected: MZT Affected: Affected: twins in remission ICD-10; depression sx (HAM-D); 16S rRNA (V3-V4 region); USEARCH 9.2 and mothur 1.38.1 α: observed OTUs, et al., MZT HR: 32 37.7 18M:53F (HAM-D, YMRS b 14); 27 BD, 45 UD; mania sx (YMRS); psychiatric and Sequencing: Illumina MiSeq, 2 × pipeline; removal of low quality Shannon Index 2019 MZT LR: 25 (CI = 8.9) MZT HR: 9M:23F outpatients; 44 medications: 22 medical hx; fasting samples (urine, 300 bp paired-end and chimeric sequences; OTU β: generalized MZT HR: 38.2 MZT LR: 5M:20F antidepressants, 15 AP, 11 blood) picking with 97% threshold; UniFrac distances (CI = 9.4) anticonvulsant, 4 lithium, 7 BZ; BMI rarified to 5244 sequences per MZT LR: 37.2 = 26.5 (SD = 7.0) sample (CI = 7.7) HR: unaffected twins with co-twin history of affective disorder; 10 BD, 22 UD; 1 on medication (not specified); BMI = 23.9 (SD = 3.1) LR: no personal of family history of affective disorder; BMI = 24.5 (SD = 3.1)

ACE = Abundance-based Coverage Estimator; AD = antidepressant medication; ADR = adverse drug reactions; ANCOVA = analysis of covariance; APSS = Attenuated Positive Symptom Syndrome; ASA24 = Automated Self-Administered 24-Hour Dietary Assessment Tool; ASRM = Altman Self-Rating Mania Scale; AOS = age of onset; AP = antipsychotic medication; BD (I/II) = bipolar disorder (type I/II); BDI = Beck depression index; BIPS = Brief Intermittent Psychotic Syndrome; BMI = Body Mass Index; bp = base pair; BPRS(−E) = Brief Psychiatric Rating Scale(−Extended); BZ = benzodiazepine; CHD = coronary heart disease; CVD = cardiovascular disease; CI = confidence interval; CIRS = Cumulative Illness Rating Scale; CPZE = chlorpromazine equivalence; DIGS = Diagnostic Interview for Genetic Studies; DOI = duration of illness; DSM-IV/V = Diagnostic and Statistical Manual of Mental Disorders - Fourth/Fifth Edition; F = female; Faith's PD = Faith's phylogenetic diversity; FEP = first-episode psychosis (all primary psychotic disorders include); FESZ = first-episode schizophrenia; GAD-7 = Generalized Anxiety Disorder 7-item scale; GRDS = Genetic Risk and Deterioration Syndrome; GAF = Global Assess- ment of Functioning; HADS = Hospital Anxiety and Depression Scale; HAM-D = Hamilton Depression Scale; HDL = high density lipoprotein; HOMA-IR = Homeostatic Model Assessment of Insulin Resistance; HR = high-risk; hs-CRP = high-sensitivity C-reactive protein; hx = history; ICD-10 = International Classification of Diseases 10th edition; IPAQ = The International Physical Activity Questionnaire; IQR = interquartile range; KEGG = Kyoto Encyclopedia of Genes and Genomes; LDL = low-density lipoproteins; LEfSe = linear discriminant analysis effect size; LR = low-risk; M = male; MS = mood stabilizer; MZT = monozygotic twins; NC = non-psychiatric comparison group; NOS = not otherwise specified; (s)OTU = (sub) operational taxonomic unit; PANSS = positive and negative syndrome score; PCoA = principal coordinates analysis; (q)PCR = (quantitative real-time) polymerase chain reaction; PD = Psychotic Disorder; PHQ-9 = Patient Health Questionnaire 9; (s)PLS-DA = (sparse) partial least-squares discriminant analysis; PSQI = Pittsburgh Sleep Quality Index; QIIME(2) = quantitative insights into microbial ecology (2); rRNA = ribosomal ribonucleic acid; SANS = Scale for the Assessment of Neg- ative Symptoms; SAPS = Scale for the Assessment of Positive Symptoms; SD = standard deviation; SF-12/36 = 12/36-Item Short-Form Health Survey; SOD = superoxide dismutase; spp. = species; SIPS = Structured Interview for Prodromal Symp- toms; SOD =;SOPS = Scale of Prodromal Syndromes; SPSS = Statistical Package for Social Sciences; SZ = schizophrenia; SZA = schizoaffective disorder; SZP = Schizophreniform Disorder; Tx = Treatment; UD = Unipolar Affective Disorder; UHR = ultra-high-risk individuals; UR = unaffected relative; WHO DDD = World Health Organization antipsychotic defined daily dose; y = years; YMRS = young mania rating scale. T.T. Nguyen et al. / Schizophrenia Research xxx (xxxx) xxx 7

Table 2 Summary of study characteristics for reports in review.

Sample characteristics Clinical sample Comparison sample

Mean (SD) Median Range Mean (SD) Median Range

Number of participants 46.5 (31.7) 35.5 12–115 42.3 (23.7) 41.0 10–77 Mean age (years) 39.6 (12.1) 41.3 20–63 37.9 (10.3) 37.9 23–55 Minimum age (years) 19.6 (5.2) 19.0 13–30 21.2 (5.3) 21.0 13–30 Maximum Age (years) 49.9 (16.4) 52.0 30–76 44.8 (16.5) 40.3 30–76 Gender ratio (M/F) 1.2 (1.0) 1.2 0–4 0.92 (0.4) 0.98 0–2 Body mass index 26.0 (3.5) 25.7 21–32 24.7 (2.7) 24.2 21–31 Age of onset (years) 22.1 (4.9) 21.5 17–28 ––– Duration of illness (years) 15.3 (14.1) 14.3 0–32 –––

F = females; M = males; SD = standard deviation.

Knowing the taxonomic composition of a microbial community tells (Schwarz et al., 2018). Similarly, order Lactobacillales was differentially only one part of the story. The research field is increasingly moving to- increased in ultra-HR compared to HR and NCs (He et al., 2018). ward understanding the functions of specific strains of these communi- Lactobacillus group was associated with increased severity of psy- ties, which offers biotechnological promise in therapeutic discovery and chotic symptoms and worse global functioning in FEP patients at the provides greater insight into the contributions of microorganisms to time of hospitalization (Schwarz et al., 2018). These studies also showed human health. This can be done by facilitating additional analysis of cur- an association between the microbiome and systemic inflammation. rent sequencing techniques and/or integrating other omics data, includ- Bifidobacterium spp. was correlated with lower serum low-density lipo- ing metatranscriptomics, metaproteomics, and metabolomics. Although protein (LDL), while Escherichia coli correlated with lower serum tri- marker gene analysis does not provide direct evidence of a community's glycerides and high-sensitivity C-reactive protein (Yuan et al., 2018). functional capabilities, PICRUSt (phylogenetic investigation of commu- Finally, one study explored possible functional pathways using PICRUSt nities by reconstruction of unobserved states) is a computational ap- to infer genetic potentials based on 16S sequences, with no differences proach that can use 16S data to predict the functional composition of in any KEGG pathways to three levels (He et al., 2018). a metagenome (Langille et al., 2013). Metagenomic sequencing can also produce detailed metabolic and functional profiles of microbial 3.2.2. Longitudinal findings communities by giving access to all genes, which can then be mapped Two studies assessed the microbiome longitudinally, tracking pa- onto known functional annotations. Metabolomics and genome-scale tient outcomes following initial hospitalization (Schwarz et al., 2018) metabolic modelling can identify bacterial metabolites and predict met- and a period of risperidone treatment (Yuan et al., 2018). FEP patients abolic pathways and products from genomic and transcriptomic data, who showed the greatest abnormalities in microbial composition from respectively (Kim et al., 2017; Ursell et al., 2014). These methods cap- NCs at hospitalization showed lower rate of disease remission at one- ture microbially produced metabolites, which can signal neighboring year follow-up, even after considering potential confounders such as microorganisms and influence host physiology, and highlight potential baseline level of global functioning, level of physical activity, BMI, dura- metabolic pathways by which gut microbes may meaningfully impact tion of antipsychotic treatment, and food intake (Schwarz et al., 2018). human health and function. Following 24 weeks of risperidone treatment, FEP patients showed increases in Bifidobacterium spp. and Escherichia coli and decreases in Clos- tridium coccoides group and Lactobacillus spp. Notably, increases in 3.2. High-risk and first episode psychosis Bifidobacterium spp. predicted increases in weight and BMI over the treatment period. Three articles met our inclusion criteria, including one of individuals at high-risk (HR) and ultra-HR for schizophrenia and two of persons 3.3. Schizophrenia with FEP. Each study used different methods for quantifying the gut microbiome, including 16S sequencing, quantitative real-time polymer- There were five articles of the gut microbiome in patients with ase chain reaction (qPCR) analyses using 16S rRNA primers, and schizophrenia. Three studies were cross-sectional investigations, and metagenomics sequencing. two studies sampled the microbiome longitudinally. All studies utilized 16S sequencing. 3.2.1. Cross-sectional findings Only one study reported community-level characteristics and found 3.3.1. Cross-sectional findings no differences in alpha-diversity among HR, ultra-HR, and NCs (He et al., All studies analyzed diversity metrics, but findings were mixed. 2018). However, beta-diversity analysis revealed that ultra-HR and HR Zheng et al. (2019) observed reduced microbial richness and evenness had significantly different global microbiome composition than NCs, in schizophrenia relative to NCs, while two other studies found no dif- with ultra-HR showing greater heterogeneity in clustering across the ferences in alpha-diversity using similar indices (Nguyen et al., 2019; principal coordinate analysis space compared to other groups. Shen et al., 2018). Nevertheless, all studies revealed beta-diversity dif- All studies reported on taxonomic differences between groups. To- ferences between schizophrenia and NC groups. gether, these investigations revealed 25 taxa that were significantly dif- All investigations reported taxonomic differences between schizo- ferent among FEP, HR groups, and NCs. Between the two studies using phrenia and NCs, although the drivers of community separation varied qPCR analysis, the single common finding was that Bacteroides spp. considerably across studies. Combining these studies, 130 taxa were sig- was not significantly different between FEP and NC groups. Yuan et al. nificantly different between schizophrenia and NCs. At the phylum (2018) reported reduced numbers of Bifidobacterium spp., Escherichia level, two studies found Proteobacteria to be different between groups; coli,andLactobacillus spp. and increased numbers of Clostridium however, it was relatively decreased in schizophrenia in one (Nguyen coccoides group in FEP compared to NCs, while Schwarz et al. (2018) et al., 2019) but relatively increased in another (Shen et al., 2018). Di- found no differences in bacterial numbers between FEP and NCs. On dif- vergent findings were also reported for genus Clostridium (Nguyen ferential abundance testing, family Lactobacillaceae was overrepre- et al., 2019; Shen et al., 2018). Six genera were relatively increased in sented among the taxa that were most strongly increased in FEP schizophrenia: Anaerococcus, Succinivibrio, Megasphaera, Collinsella,

Table 3 Studies of the microbiome in individuals at high-risk for and with ﬁrst episode psychosis.

Publication Diversity patterns Taxonomic differencesa Association with clinical Functional potential Longitudinal changes Limitations features

He et al., α: no difference Orders: UHR ↑ N/A No differences in KEGG N/A − Small sample size for 2018 between HR, UHR, Clostridiales, pathways into 3 levels; UHR group and NC (observed Lactobacillales, and UHR ↑ acetyl coenzyme − Did not examine rela- OTUs, Shannon Bacteroidales compared A synthesis pathway tionship of community Index) to HR and NC compared to HR and diversity/specific taxa β: HR and UHR Genera: UHR ↑ NC using KEGG to demographic or different from NC Lactobacillus and Orthology database clinical/disease-- (PCoA) with Prevotella compared to specific features in increased HR and NC groups heterogeneity in Species: UHR ↑ − No data on BMI or diet clustering in UHR Lactobacillus ruminis − Functional profiling compared to HR compared to HR and NC performed on the and NC (PLS-DA) basis of 16S rRNA gene data is limited; only inferences can be made about the genome corresponding to the specific marker gene sequence, as well as about the functional potential of a given genome Schwarz N/A qPCR: numbers of Lachnospiraceae, Bacteroides N/A Microbiota clustering at − Small sample size et al., bacteria not different spp., Lactobacillus intake associated with ↑ − No community-level 2018 between FEP and NC correlated with ↑ psychotic remission at 12-month characteristics Metagenomic symptoms. Lachnospiraceae, follow-up; 70% FEP that reported (alpha- and sequencing: Families: Bacteroides spp., and clustered with NCs showed beta-diversity) FEP ↑ Lactobacillaceae, predominant bacteria remission, compared to − qPCR analysis limited Halothiobacillaceae, associated with ↑ negative only 28% of patients with to only 5 bacterial Brucellaceae, and symptoms. Lactobacillus “abnormal,” even after groups Micrococcineae, ↓ correlated with ↑ positive controlling for baseline GAF − Model predicting Veillonellaceae symptoms. remission only used Genera: FEP ↑ Ruminococcaceae, top 5 families rather Lactobacillus, Bacteroides spp., than the entire popu- Tropheryma, Lactobacillus, and lation Halothiobacillus, predominant bacteria − More specific informa- Saccharophagus, associated with ↓ GAF. tion about/- Ochrobactrum, Duration of AP treatment examination of the Deferribacter, and not correlated with impact of AP medica- Halorubrum, ↓ Anabaena, bacteria. Subgroup of FEP tion use Nitrosospira, and with greatest differences in − Further examination Gallionella composition from NCs of relationship with showed ↑ negative metabolic and inflam- symptoms and ↓ GAF but matory biomarkers not with positive collected symptoms. − No information regarding diet − No analyses of functional potential Yuan et al., N/A FESZ ↓ At baseline, Bifidobacterium N/A After 24 weeks of − Single arm study: lack 2018 Bifidobacteriumspp., spp. correlated with ↓ risperidone treatment, ↑ of non-treatment con- Escherichia coli, serum LDL; Escherichia coli Bifidobacterium spp. and trol group, no follow-- Lactobacillusspp., and ↑ correlated with ↓ serum Escherichia coli and ↓ up in NC Clostridium coccoides triglycerides and hs-CRP, Clostridium coccoides group − No community-level group. No difference in after controlling for age, and Lactobacillus spp. No characteristics Bacteroidesspp. between gender, smoking status, and change in Bacteroides spp. reported (alpha- and groups DOI Hierarchical multiple linear beta-diversity) regression analysis shows − Limited microbiome that only △Bifidobacterium investigation to qPCR spp. was correlated with analysis of only 5 △weight over 24 weeks, bacteria, rather than after controlling for age, performing OTU gender, smoking status, and analysis DOI. No other relationships between changes in fecal bacteria and changes in metabolic parameters.

AP = antipsychotic medication; DOI = duration of illness; F = female; FEP = ﬁrst-episode psychosis (all primary psychotic disorders include); FESZ = ﬁrst-episode schizophrenia; GAF = Global Assessment of Functioning; HR = high-risk individuals; hs-CRP = high-sensitivity C-reactive protein; KEGG = Kyoto Encyclopedia of Genes and Genomes; LDL = low-density lipoproteins; M = male; NC = non-psychiatric comparison group; OTU = operational taxonomic unit; PCoA = principal coordinates analysis; PLS-DA = partial least-squares discriminant analysis; qPCR = quantitative real-time polymerase chain reaction; rRNA = ribosomal ribonucelic acid; SD = standard deviation; spp. = species, SZ = schizophrenia; UHR = ultra-high- risk individuals. a ↑↓ arrows indicate increase or decrease in relative abundance, when referring to taxonomic differences.

Klebsiella, Methanobrevibacter,whilefive genera were decreased in between concordant and discordant pairs (Vinberg et al., 2019). Two ar- schizophrenia: Haemophilus, Sutterella, Blautia, Coprococcus,and ticles reported no differences between BD and NCs or their unaffected Roseburia. At the family level, Veillonellaceae, Prevotellaceae, first-degree relatives (Coello et al., 2019; Painold et al., 2019). Differ- Bacteroidaceae,andCoriobacteriaceae were increased in schizophrenia, ences in alpha-diversity within BD subgroups suggested that diversity whereas Lachnospiraceae, Ruminococcaceae, Norank,andEnterobacteria- patterns may vary depending on mood state and gender; species even- ceae were decreased. One study evaluated the specificity of its findings ness was reduced in depressed BD compared to euthymic BD in schizophrenia, comparing results to a previous study of patients (Bengesser et al., 2019), and species richness was decreased in with major depressive disorder (Zheng et al., 2016), and found that women with schizophrenia or BD who were treated with antipsychotic only a minority of taxa overlapped, suggesting that schizophrenia has medications, compared to those not-treated with antipsychotics a somewhat distinct microbial signature. (Flowers et al., 2019). With regards to beta-diversity differences, reports Psychosis symptom severity was positively correlated with were mixed. One study found that community membership differed be- Bacteroidaceae, Streptococcaceae,andLachnospiracea and negatively tween BD and NCs (Coello et al., 2019), while another investigation with Veillonellaceae (Zheng et al., 2019). Findings for Ruminococcaceae using the same measure did not (Painold et al., 2019). Likewise, one ar- were mixed. One study observed two different operational taxonomic ticle revealed community structure differences (Evans et al., 2017), units correlated with psychosis symptoms severity, one positively and while two investigations found no differences (Coello et al., 2019; the other negatively (Zheng et al., 2019). Another study reported Painold et al., 2019). Beta-diversity was also not different among mono- Ruminococcaceae to be inversely associated with severity of negative zygotic twins concordant or discordant for affective disorders and those symptoms (Nguyen et al., 2019). Bacteroides was positively related to without affective disorders (Vinberg et al., 2019). depressive symptoms, and Verrucomicrobia with self-reported mental Combining these studies, seven taxa were different between BD and well-being. These studies also reported relationships with other clinical NCs. At the phylum level, only Actinobacteria was found to be signifi- and health variables. Cyanobacteria correlated with increased age of cantly different between groups, with abundance relatively increased onset, Coprococcus with greater risk for developing coronary heart dis- in BD (Painold et al., 2019). At the family and genus levels, two studies ease, and Actinobacteria with greater number of years of smoking revealed Ruminococcaceae and Faecalibacterium to be relatively de- (Nguyen et al., 2019). creased in BD (Evans et al., 2017; Painold et al., 2019). Other signifi- Finally, several functional metabolic pathways differed between cantly different taxa included Coriobacteriaceae and Flavonifractor, schizophrenia and NCs, including vitamin B6, fatty acid, starch and which were increased in BD, and Christensenellaceae,whichwasde- sucrose, tryptophan, cysteine, methionine, and linoleic acid metabo- creased. Neither Bifidobacterium nor Lactobacillus was found to be differ- lism, as well as degradation of some xenobiotics (e.g., foreign sub- ent between BD and NCs (Aizawa et al., 2018). stances to the body or ecological system) (Shen et al., 2018). All these studies examined the relationship of the gut microbiome Specific taxa were associated with these differential metabolic path- with clinical variables, including disease severity, psychiatric symptoms, ways: Blautia, Coprococcus,andRoseburia were negatively associ- medication use, and health variables. Longer duration of illness was as- ated with vitamin B6, taurine, hypotaurine metabolic pathways sociated with decreased alpha-diversity (Painold et al., 2019). Depres- and positively associated with the methane metabolic pathway sion symptom severity was positively correlated with (Shen et al., 2018). Enterobacteriaceae and negatively with Faecalibacterium, Clostridiaceae and Roseburia (Painold et al., 2019), whereas sleep was positively asso- 3.3.2. Longitudinal findings ciated with Faecalibacterium (Evans et al., 2017). Atypical antipsychotic Two studies evaluated changes in the gut microbiome following pro- treatment was associated with reduced gut biodiversity, particularly in biotic and prebiotic administration. Nagamine et al. (2018) explored women (Flowers et al., 2019, 2017). Patients on atypical antipsychotics whether 6-month supplementation of prebiotic 4G-β-D- had relatively increased levels of Lachnospiraceae, while non-treated in- galactosylsucrose, an oligosaccharide that is selectively utilized by dividuals had preferentially higher levels of Akkermansia and Alistipes. Bifidobacterium and has been reported to be beneficial in chronic in- Smoking correlated with increased presence of Flavonifractor (Coello flammatory bowel disease (Teramoto et al., 1996), could improve low et al., 2019), and better self-reported physical health was associated body weight in schizophrenia. They observed that Bifidobacterium was with increased Faecalibacterium, Anaerostipes and Ruminococcaceae increased and Clostridium subcluster XIVa was reduced. Changes in mi- and decreased Enterobacteriaceae (Evans et al., 2017). BD patients crobial composition were also accompanied by significant increases in with higher BMI and metabolic syndrome showed increased weight and BMI. On the other hand, there were no changes in Lactobacillaceae, Lactobacillus,andCoriobacteriaceae (Painold et al., Bifidobacterium following 4-week administration of probiotic 2019). Microbiome composition was also related to serum inflamma- Bifidobacterium breve A-1, despite improved depression and anxiety tory and metabolic biomarkers (Painold et al., 2019). BD with higher symptoms (Okubo et al., 2019). When the baseline microbial composi- IL-6 had increased Lactobacillales, Lactobacillaceae, Lactobacillus, tions were evaluated, there was no significant difference in alpha- Streptococcaceae, and Streptococcus, compared to BD with lower IL-6. diversity, beta-diversity, or relative abundances at the phylum level be- Similarly, higher total cholesterol was associated with increased tween treatment responders compared to non-responders; at the genus Clostridiaceae and lower LDL was associated with increased level, treatment non-responders had higher levels of Parabacteroides. Prevotellaceae and Prevotella. Increased thiobarbituric acid reactive substances, a parameter of oxidative stress, were associated with increased 3.4. Bipolar disorder Eubacterium, and higher tryptophan levels were associated with Lacto- bacillus, Lactobacillaceae, Coriobacteriaceae,andClostridiaceae. Eight articles investigated the gut microbiome in BD. All studies were cross-sectional except for one longitudinal investigation, which 3.4.2. Longitudinal findings explored intestinal microbial changes following prebiotic administra- One article evaluated changes in the gut microbiome in patients tion. Seven studies utilized 16S sequencing, and one utilized qPCR (ex- with schizophrenia and BD following administration of resistant starch amining only Bifidobacterium and Lactobacillus subgroups). (Flowers et al., 2019), based on evidence from prior studies that have shown an inverse association between diets consisting of resistant 3.4.1. Cross-sectional findings starch and occurrence of obesity and diabetes mellitus in the general Six articles reported on measures of alpha- and beta-diversity. population (Higgins et al., 2004; Johnston et al., 2010). Nondigestible Monozygotic twins concordant for affective disorders had reduced spe- plant fibers, such as resistant starches, are selectively fermented by bac- cies richness compared to unaffected twins, but richness did not differ terial species in the large intestine and lead to the production of short-

laect hsatcea:TT gyn .Htaa,T ocoe,e l,Gtmcoim nsrosmna lnse:Asseai eiwand review systematic A illnesses: mental serious in microbiome Gut al., et Kosciolek, T. Hathaway, H. Research, Schizophrenia Nguyen, evaluation, T.T. critical as: article this cite Please Table 4 Studies of the microbiome in patients with schizophrenia.

Publication Diversity patterns Taxonomic differencesa Association with clinical Functional potential Longitudinal changes Limitations features

Nagamine N/A N/A N/A N/A After 6 months of treatment with − Small sample size et al., 4G-β-D-galactosylsucrose − Lack of non-treatment control 2018 (prebiotic treatment for group, cannot exclude a placebo underweight patients), ↑ effect Bifidobacterium and ↓ Clostridium − No community-level character- subcluster XIVa, which was istics reported (alpha and beta-- accompanied by increase in diversity) weight/BMI in underweight SZ − Further investigation of whether gut microbiome composition could predict changes in weight/BMI and other targeted outcomes Nguyen α: no difference between Phyla: SZ ↓ Proteobacteria Cyanobacteria correlated with ↑ age N/A N/A − Small sample size https://doi.org/10.1016/j.schres.2019.08.026 et al., SZ and NC (observed Genera: SZ ↑ Anaerococcus, and ↓ of onset; Bacteroides associated with − Did not assess relationship with 2019 OTUs, Shannon Index, Haemophilus, Sutterella, Clostridium ↑ depressive symptoms; AP medications ↓ − Faith's PD) Across all taxonomic levels: 35 OTUs Ruminococcaceae correlated with Heterogeneous sample includ- xxx (xxxx) xxx Research Schizophrenia / al. et Nguyen T.T. β: different between SZ different between SZ and NC (33 negative psychosis symptoms; ing patients with SZA and some and NC (unweighted order Clostridiales, 1 class Coprococcus associated with ↑ chronic disease UniFrac, Bray-Curtis Gammaproteobacteria, 1 class Framingham CHD risk; dissimilarity), with Erysipelotrichi) Verrucomicrobia correlated with ↑ tighter clustering in NC mental well-being; Actinobacteria than SZ (unweighted correlated with ↑ greater number of UniFrac, Bray-Curtis years of smoking in SZ dissimilarity) Okubo α: at baseline, no Phlyla: at baseline, no differences in N/A N/A No change in the genus − Open-label, single arm study: et al., difference between relative abundances between Bifidobacterium after 4 weeks of lack of non-treatment control 2019 responders and responders and non-responders Bifidobacterium breve A-1 treatment group, cannot exclude a placebo non-responders Genera: at baseline, non-responders or 8 weeks (post-observation), effect (Shannon Index) ↑ Parabacteroides although anxiety and depressive − Further investigation of whether β: no difference between symptoms were improved gut microbiome composition responders and could predict changes in anxiety non-responders (UniFrac and depressive symptoms and distances) cytokines Shen et al., α: no difference between Phyla: SZ ↑ Proteobacteria N/A Vitamin B6, fatty acid, starch and N/A − Excluded patients with chronic 2018 SZ and NC (number of Genera: SZ ↑ Succinivibrio, sucrose, tryptophan, cysteine, disease, which is less represen- reads, Faith's PD, Megasphaera, Collinsella, Clostridium, methionine, and linoleic acid tative of the SZ population observed OTUs, Shannon Klebsiella, Methanobrevibacter;NC↑ metabolism; and degradation of − Excluded patients with SZA and Index, Simpson, ACE, Blautia, Coprococcus, and xenobiotics different between SZ other SZ spectrum disorders Chao1) Roseburiawere and NC. Blautia, Coprococcus, and − No indication of whether sam- β: different between SZ Species: SZ ↑ Collinsella aerofaciens, Roseburia associated with ↓ vitamin ples were matched on demo- and NC (unweighted Bacteroides fragilis, and ↓ Roseburia B6, taurine, hypotaurine metabolic graphic variables UniFrac, PCoA), with faecis, Blautia producta, Collinsella pathways and ↑ methane metabolic − Validation of a microbiome-- tighter clustering in NC plebeius pathway based schizophrenia classifier than SZ (unweighted based on small sub-sample UniFrac) − Did not examine relationship of community diversity/specific taxa to demographic or clinical/disease-specific features in groups − Functional profiling performed on the basis of 16S rRNA gene data is limited; only inferences can be made about the genome corresponding to the specific marker gene sequence, as well T.T. Nguyen et al. / Schizophrenia Research xxx (xxxx) xxx 11

chain fatty acids (SCFA) that have the potential to improve health (Louis et al., 2007; Nugent, 2005). One particular SCFA, butyrate, has been found to be associated with improved depression-related behaviors in a mouse model by increasing serotonin concentration and brain- derived neurotrophic factor (BDNF) expression as well as restoring blood-brain barrier impairments (Sun et al., 2016). Although alpha- diversity remained unchanged, beta-diversity differences were observed following 14-day supplementation of resistant starch. Specifi- cally, the relative abundances of phylum Actinobacteria increased, Uneven read lengths between sequences (arbitrary trimming of sequences) might create systematic biases in downstream analyses Differential abundance testing performed only on the basissupervised of method (LefSe) and no unsupervised tests (for beta-diversity and differential abundance) reported. Likely over-estimating the observed effects as about the functional potential of a given genome genera Bacteroides and Parabacteroides decreased, and resistant − − starch–degrading species Bifidobacterium faecale and Bifidobacterium adolescentis increased. rational taxonomic unit; rRNA = ribosomal ri-

4. Discussion

The microbiome revolution has opened new frontiers for examining host-microbe associations in the context of understanding brain and behavioral health and better conceptualizing psychiatric disorders. In- deed, there have been more conceptual reviews and opinion papers on the gut-brain axis and its potential role in psychiatric disorders than there are empirical, data-driven investigations examining the com-

uration of illness; F = female; Faith's PD = Faith's phylogenetic diversity; LefSe = position of the microbiota in SMI and its role in disease presentation and treatment. This paper provides a detailed summary of the evidence regarding alterations of the gut microbiome in people with SMI, including

east-squares discriminant analysis; OTU = ope FEP, schizophrenia, and BD, building upon our previous study (Nguyen et al., 2018b). We did not ﬁnd any other published paper that systemat- ically reviewed the available data on the gut microbiome in SMI. In our previous review, there were ﬁve studies, only three of which were fo- cused on the gut microbiome. Since then, in just under two years, 13 additional papers on the gut microbiome have been published. Although multiple papers were published from the same groups of investigators,

N/A N/A presumably using overlapping participant cohorts, the studies reviewed here represent a broader range of patient types and reﬂect individuals ↑

, and with a wider geographical spread across the US, Europe, and Asia, com- ↓ pared to our earlier article. All the reviewed studies found alterations of the gut microbiome in psychosis

↓ patients with SMI compared to NCs. However, speciﬁc microbial metrics Ruminococcaceae OTUs correlated

and reported to be anomalous across articles varied, and there was minimal ↑ Streptococcaceae correlated with , consensus with regards to microbial diversity patterns, relative abundance, or directionality of differences in taxa. Even within specific patient populations, there were more differences in findings across investigations than similarities. The most consistent finding was Lachnospiraceae symptom severity. psychosis symptom severity; Bacteroidaceae 2 psychosis symptom severity. 2 different OTUs from correlated with Veillonellaceae among studies of schizophrenia, all of which reported differences in otic medication; BMI = body mass index; CHD = coronary heart disease; DOI = d

, beta-diversity between patients and NCs. Yet, when looking at the tax- . parison group; PCoA = principal coordinates analysis; PLS-DA = partial l

SZ: onomic drivers of these global community differences, 130 taxa were , , ↑ (4 Norank found to be different across ﬁve studies, with little consistency across investigations. Two studies revealed Ruminococcaceae and (16 OTUs),

Bacteroidaceae Faecalibacterium to be relatively decreased in BD. Abundance of Rikenellaceae , were similarly up- Brucellaceae (2 OTUs); 54 OTUs (12 OTUs), , Aerococcaceae

, Ruminococcaceae was also reported across several investigations of (3 OTUs), (5 OTUs), , and (4 OTUs), schizophrenia and BD to be associated with better clinical characteris- ﬁ Enterobacteriaceae tics, including reduced psychosis symptom severity (and speci cally abundance, when referring to taxonomic differences. negative symptoms) and improved self-reported physical health Lachnospiraceae Veillonellaceae

dobacteriaceae (Evans et al., 2017; Nguyen et al., 2019; Zheng et al., 2019), suggesting relative ﬁ SZ: OTUs); Families most predictive of SZ diagnosis, based on stepwise regression, were Bi Ruminococcaceae (5 OTUs), Families: 77 differential OTUs between SZ and NC; 23 OTUs Veillonellaceae Prevotellaceae Bacteroidaceae Coriobacteriaceae ↓ Pasteurellaceae Compared to NC, only Ruminococcaceae Acidaminococcaceae and or down-regulated in SZ and UD that this taxon may be protective. Another notable ﬁnding was that lactic acid bacteria (i.e., order Lactobacillales, family Lactobacillaceae, genus Lactobacillus) were rela- SZ

↓ tively more abundant in SMI and associated with worse outcomes, including increased severity of psychotic symptoms, poorer global functioning (He et al., 2018; Schwarz et al., 2018), higher BMI and metabolic syndrome, and higher serum levels of pro-inﬂammatory IL-6 (Painold et al., 2019). These results are consistent with prior studies of : microbial richness : different between SZ β and NC (PLS-DA) from order to species levels; results not clustered by sex or AP use or type α (Chao) and diversity (Shannon Index) the oropharyngeal microbiome, in which Lactobacillus gasseri,along with a bacteriophage that preferentially infects Lactobacillus gasseri,

arrows indicate increase or decrease in were relatively increased in schizophrenia (Castro-Nallar et al., 2015; et al., 2019 ↑↓ Yolken et al., 2015). Transmission of microbes from the mouth to the Zheng a bonucleic acid; SD = standard deviation; SZ = schizophrenia; SZA = schizoaffective disorder; Tx = treatment; UD = unipolar affective disorder. linear discriminant analysis effect size; M = male; NC = non-psychiatric com ACE = Abundance-based Coverage Estimator; AOS = age of onset; AP = antipsych gut is a constant process, and oral-fecal transmission is an important

laect hsatcea:TT gyn .Htaa,T ocoe,e l,Gtmcoim nsrosmna lnse:Asseai eiwand review systematic A illnesses: mental serious in microbiome Gut al., et Kosciolek, T. Hathaway, H. Research, Schizophrenia Nguyen, evaluation, T.T. critical as: article this cite Please Table 5 Studies of the microbiome in patients with bipolar disorder.

Publication Diversity patterns Taxonomic differencesa Association with clinical Functional Longitudinal changes Limitations features potential

Aizawa N/A No difference in Bifidobacterium or Bifidobacterium and Lactobacillus counts N/A N/A − qPCR analysis limited to only 2 et al., Lactobacillus counts between BD and NC, even correlated with ↓ sleep symptoms. No bacterial groups 2018 when BDI/BDII patients and males/females association with overall depressive or manic − Mixed sample of BD (different examined separately and controlling for BMI. symptoms. Bifidobacterium counts associated sub-diagnoses, mood states) ↓ with cortisol levels. No relationship between may have diluted findings bacteria counts and dose of AP or between − Did not account for length/- those on/off MS. type of medication use Bengesser α: BD depressed ↓ evenness N/A Methylation of cg05733463 of the clock gene N/A N/A − Small sample size et al., (Simpson Index) compared to BD ARNTL correlated with ↓ alpha-diversity (both − Did not report on beta-- 2019 euthymic richness and evenness) in BD. Neither diversity or taxonomic differ- depressive nor manic symptoms were ences associated with any measure of − Did not examine patients in alpha-diversity. manic or mixed states https://doi.org/10.1016/j.schres.2019.08.026 − Did not account for medication use or diet Coello α: no difference among BD, UR, and BD ↑ Flavonifractor compared to UR and NC, Flavonifractor associated with smoking and N/A N/A − Most BD and UR were ..Nue ta./ShzprnaRsac x xx)xxx (xxxx) xxx Research Schizophrenia / al. et Nguyen T.T. et al., NC (observed OTUs, Shannon Index) even after controlling for age, sex, smoking, female sex, but not age, waist circumference, smokers, although statistically 2019 β: unweighted UniFrac different waist circumference, and physical activity; physical activity, DOI, depressive and manic controlled for this variable, between BD and NC (but not Non-smoking BD ↑ Flavonifractor compared to symptoms, medication, and hs-CRP; no cannot rule out potential con- weighted UniFrac); weighted non-smoking NC difference in Flavonifractor among BD subtype founding effect UniFrac different between BD and or mood states − Mood states evaluated as a UR; no difference between UR and dichotomous variable NC (unweighted and weighted (euthymic vs. affective state), UniFrac) rather than individual states, which may have diluted findings − No data on diet Evans β: different between BD and NC (Yue BD ↓ Faecalibacterium and an unclassified Faecalibacterium associated with ↑ physical N/A N/A − Did not report alpha-diversity et al., and Clayton distance) member from the Ruminococcaceae family health, ↓ depression, ↑ sleep quality; − Requires further examination 2017 Anaerostipes and Ruminococcaceae associated of associations with disease-- with ↑ physical health, Enterobacteriaceae specific features ↓ associated with physical health. (e.g., duration of illness, number of mood episodes) − Did not quantify the amount of or investigate the effects of medication use among BD − Used mean scores on self-- report scales rather than assessment(s) most tempo- rally related to time of stool collection Flowers α: AP-treated women ↓ diversity AP-treated patients ↑ Lachnospiraceae Akkermansia ↓ in non-obese AP-treated N/A N/A − Further examination of rela- et al., (Simpson Diversity Index); no Non-AP-treated patients ↑ Akkermansia and patients. tionship with duration of ill- 2017 differences between BD men Sutterella ness or other indicators of β: different between BD medication disease or symptom severity groups (Yue and Clayton distance). − Further investigation of rela- fi OTUs identi ed for differing tionship of comorbid medical directions of the medication groups: conditions or other metabolic Lachnospiraceae, Alistipes, and biomarkers on microbiome, Akkermansia. given the relationship between atypical APs and metabolic disease − No information regarding diet, which is an important environmental factor that drives laect hsatcea:TT gyn .Htaa,T ocoe,e l,Gtmcoim nsrosmna lnse:Asseai eiwand review systematic A illnesses: mental serious in microbiome Gut al., et Kosciolek, T. Hathaway, H. Research, Schizophrenia Nguyen, evaluation, T.T. critical as: article this cite Please gut microbial composition and as atypical APs could increase appetite Flowers α: at baseline, no difference between non-AP ↑ Alistipes compared to AP group AP-treated women ↓ alpha-diversity (Simpson N/A Following 14-day prebiotic (resistant starch) − Single arm study: lack of et al., AP and non-AP groups (Inverse Diversity Index); no differences between men. administration, no changes in alpha-diversity; non-treatment control group, 2019 Simpson Index) significant beta-diversity differences no follow-up in non-AP β: at baseline, no difference between (Bray-Curtis); ↑ phylum Actinobacteria and ↓ treated group AP and non-AP groups (Bray-Curtis) genera Bacteroides and Parabacteroides. − No data on adherence, which ↑ – 4.2-fold resistant starch degrading species may contribute to variability Bifidobacterium faecale and Bifidobacterium in results adolescentis. − Differences in represented No changes to phyla Firmicutes, Bacteroidetes, diagnoses within the AP and Proteobacteria; other Bifidobacteria; Ruminococcus bromii (resistant group, with increased num- starch–degrading species); or Faecalibacterium bers of SZ and SZA patients prausnitzii, Eubacterium rectale, and − Further analyses between Eubacterium hallii (butyrate-producing diagnostic groups (SZ vs. BD) species) Painold α: no difference between BD and NC Phyla: BD ↑ Actinobacteria, compared to NC Alpha-diversity (observed OTUs) correlated N/A N/A − BD had significant higher BMI ↑ ↓ https://doi.org/10.1016/j.schres.2019.08.026 et al., (observed OTUs, Chao1, Shannon Class: BD Coriobacteria with DOI. No relationship between values; although analyses 2019 Index, Simpson Index), even after Order: BD ↑ Coriobacteriales alpha-diversity and serum concentrations of were repeated with severely patients with diabetes and severe Family: BD ↑ Coriobacteriaceae and ↓ CRP, IL-6, total cholesterol, HDL, LDL, TRP, KYN, obese patients excluded, sub- N xxx (xxxx) xxx Research Schizophrenia / al. et Nguyen T.T. obesity (BMI 35) excluded Ruminococcaceae oxidative stress parameters, depression levels, sample size is considerably β ↓ : no difference between BD and NC Genus: BD Faecalibacterium or anthropometric measurements. Sample smaller (unweighted and weighted UniFrac), median split of BD with high/low values of − BD patients acutely ill and on even after excluding patients with serum biomarkers: BD with high IL-6 ↑ different medications, which diabetes and severe obesity Lactobacillales, Lactobacillaceae, Lactobacillus, may have contributed to vari- Streptococcaceae, and Streptococcus compared to BD with lower IL-6; high total cholesterol ↑ ability − Clostridiaceae; low LDL cholesterol ↑ Did not examine or account Prevotellaceae and Prevotella. No difference for medication differences between high and low HDL groups. High TRP differed significantly in the Lactobacillus, Lactobacillaceae, Coriobacteriaceae and Clostridiaceae; no difference in KYN; high TBARS ↑ Eubacterium; low MDA ↑ Faecalibacterium. BD with high BMI ↑ Lactobacillus, Lactobacillaceae, and Bacilli. BD with metabolic syndrome ↑ Lactobacillaceae, Lactobacillus, and Coriobacteriaceae. BD with clinically relevant depressive symptoms (BDI ≥ 18) ↑ Enterobacteriaceae and ↓ Clostridiaceae and Roseburia Vinberg α: difference between affected, HR, At phylum, class, order, family, and genus Christensenellaceae not associated with N/A N/A − Unclear if lower et al., and LR MZT (observed OTUs); LR ↑ levels, unclassified Firmicutes was only difference in depressive symptoms between Christensenellaceae due to 2019 richness than affected group; no predictive taxa of ‘disease.’ MZT decreased relative abundance difference between affected and HR At OTU level, single OTU Christensenellaceae ↓ or detection limitations β : no difference between affected, in affected and HR compared to LR. − Uneven group sizes ↓ HR, and LR (generalized UniFrac) Christensenellaceae in UD affected/HR and BD − Affected and HR MZT smoked affected/HR, compared to LR. In only had higher prevalence of discordant MZT, no differences in smoking and higher BMI; LR Christensenellaceae between affected and HR had higher rates of alcohol when looking at UD, BD, or both combined. consumption − Further investigation of medication use in affected group

AOS = age of onset; AP = antipsychotic medication; ARNTL = Aryl hydrocarbon receptor nuclear translocator-like protein 1; BD (I/II) = bipolar disorder (type I/II); BDI = Beck Depression Index; BMI = body mass index; CHD = coronary heart disease; CRP = C-reactive protein; DOI = duration of illness; F = female; Faith's PD = Faith's phylogenetic diversity; HDL = high-density lipoprotein; hs-CRP = high sensitivity C-reactive protein; IL = interleukin; KYN = kynurenine; LDL = low- density lipoprotein; LR = low risk; M = male; MDA = malondialdehyde; MS = mood stabilizer; MZT = monozygotic twin; NC = non-psychiatric comparison group; qPCR = quantitative real-time polymerase chain reaction; rRNA = ribosomal ribonucleic acid; SD = standard deviation; SZ = schizophrenia; SZA = schizoaffective disorder; TBARS = thiobarbituric acid reactive substances; TRP = tryptophan; UD = unipolar affective disorder; UR = unaffected relative. a ↑↓ arrows indicate increase or decrease in relative abundance, when referring to taxonomic differences. 13 14 T.T. Nguyen et al. / Schizophrenia Research xxx (xxxx) xxx

with certain chronic diseases, including hypertension and diabetes, that may affect the stability of the gut microbiota, while Nguyen et al. (2019) did not exclude patients with these diseases, as physiological and metabolic changes are inherent to the disorder and its associated lifestyle (Mitchell et al., 2011). Articles on BD tended to be most heterogeneous in terms of patient samples, with many studies including patients with various subtypes and in different mood states. Indeed, alpha-diversity was decreased in depressed compared to euthymic patients, suggesting that diversity patterns may also vary depending on mood state (Bengesser et al., 2019). The potential clinical implications of gut microbiome research are exciting. Despite decades of pharmacological treatment development, many patients with SMI do not respond to medications and experience many adverse side effects (Lally and MacCabe, 2015). The available medications reduce symptoms and prevent relapse but do not affect the underlying etiopathology. Modulation of the gut microbiota may be a tractable strategy for developing novel treatments without major side effects or high costs. Probiotic bacteria may exert health- promoting effects through various mechanisms that may be strain- specific (e.g., lactic acid bacteria) or widespread across a diversity of Fig. 2. Schematic representation of alpha- and beta-diversity. Three samples from different strains, including normalization of perturbed microbiota, inhibition of populations are present, each containing different numbers and proportions of taxa (blue triangles, green squares, and red stars). Alpha-diversity quantifies microbial diversity potential pathogens, production of useful metabolites or enzymes, and within a sample by characterizing species number, distribution, etc. and represents immunomodulation (Hill et al., 2014). Trials of probiotics have reported within-sample diversity. Hence, each sample receives a single numeric value to improve symptoms of depression and psychological distress in pa- α representing its -diversity (i.e., univariate variable); the exact numeric values will de- tients with major depression (Akkasheh et al., 2016; Chahwan et al., pend on the metric used (e.g. Shannon diversity, number of observed features, or Faith's phylogenetic diversity). Beta-diversity represents between-sample diversity and com- 2019), multiple sclerosis (Kouchaki et al., 2017), chronic fatigue syn- pares microbial dissimilarity between each pair of samples. In this case, each pair of sam- drome (Rao et al., 2009), and irritable bowel syndrome (Pinto- ples has a numeric value representing their β-diversity (i.e., multivariate variable), which Sanchez et al., 2017). However, the findings are far from conclusive may be represented as a matrix of values. Similar to alpha-diversity, the exact numeric and there is very limited evidence for the efficacy of probiotic or prebi- values will depend on the metric used (e.g., Bray-Curtis dissimilarity, or UniFrac [phyloge- otic interventions on changing microbial composition in SMI (Okubo netic] distance). et al., 2019). Conversely, prebiotics led to changes in microbial composition but without changes in mood or psychiatric symptoms (Flowers factor that can determine gut composition (Schmidt et al., 2019). The in- et al., 2019; Nagamine et al., 2018). These investigations shared a creased presence of lactic acid bacteria in this population and its associ- major weakness in that they were all open-label, single-arm treatment ation with worse clinical, inflammatory, and metabolic profiles is studies without a non-treatment control group. Thus, we cannot deter- surprising given that lactic acid bacteria are often considered health- mine whether observed microbial changes are due to treatment or promoting and anti-inflammatory (Ménard et al., 2004). These taxa other potential sources of bias. Earlier randomized, double-blinded, are commonly found in the composition of probiotics (Naidu et al., placebo-controlled trials investigating probiotic interventions in SMI 1999) and have been shown to enhance immunity following supple- have reported to improve bowel function (Dickerson et al., 2014), re- mentation (Schiffrin et al., 1997; Sheih et al., 2001). Schizophrenia and duce levels of intestinal inflammatory indices (Severance et al., 2017), BD have been associated with chronic inflammatory states (Berk et al., and increase levels of systemic immunomodulatory proteins (Tomasik 2011; Kirkpatrick and Miller, 2013), so reduced abundance of these et al., 2015), but the findings with regards to psychiatric outcomes are taxa might have been expected. Findings of lactic acid bacteria among still mixed (Dickerson et al., 2018, 2014; Severance et al., 2017). These other psychiatric and neurological disorders have also been mixed, studies did not assess microbial biomarkers pre- or post-treatment, so with lower levels reported in major depressive disorder (Aizawa et al., the relationship of the immune and clinical responses to gut microbial 2016) and increased levels in autism spectrum disorders (Tomova composition remains unclear. et al., 2015). In light of significant discrepancies, particularly as they relate to Although findings hint at the possibility that psychotropic medica- differences and directionality of specific taxa, we find it important tions may impact gut microbial composition, the data are still too lim- to highlight the issue of compositionality. All microbiome data are ited to draw definitive conclusions. Only six studies examined compositional, meaning that they are relative and, on their own, antipsychotic treatment. Two cross-sectional studies, reporting on over- carry no information about absolute abundances, regardless of nor- lapping cohort of patients, showed that atypical antipsychotic treat- malization steps taken (e.g., rarefaction or differential expression ment may be associated with reduced gut biodiversity, particularly in analysis, such as DESeq) (Gloor et al., 2017). Relevant statistical women (Flowers et al., 2019, 2017). Another longitudinal study found methods that account for the compositional nature of the data that risperidone treatment may result in increased Bifidobacterium (e.g., based on log-ratios) must be used, otherwise large numbers spp. and Escherichia coli and decreased in Clostridium coccoides group of false-positive data are likely (Knight et al., 2018; Mandal et al., and Lactobacillus spp. (Yuan et al., 2018). Three articles did not find 2015; Weiss et al., 2017). Also, this calls for caution in language de- any association between microbial composition or taxa with antipsy- termining the directionality of differential abundance changes. For chotic dose or treatment duration (Aizawa et al., 2018; Nguyen et al., example, it may be misleading to state that a genus of bacteria has 2018a, 2018b; Schwarz et al., 2018). increased without specifying a point of reference (i.e., another Discrepancies across articles may be explained, at least partly, by genus) in relation to which this has happened (Christensen et al., heterogeneity in study designs, methodologies, and study participant 2009; Morton et al., 2019, 2017). Thus, the compositionality problem characteristics. Although a strength of some investigations included may partially explain inconsistencies and sparse overlap in the re- strict within-sample classification and characterization, between- sults of specific microbial investigations. sample variability across studies likely contributed to differences in The studies discussed represent a variety of experimental ap- comparability. For example, Shen et al. (2018) excluded participants proaches to analyze the microbiome (e.g., 16S rRNA marker gene

Please cite this article as: T.T. Nguyen, H. Hathaway, T. Kosciolek, et al., Gut microbiome in serious mental illnesses: A systematic review and critical evaluation, Schizophrenia Research, https://doi.org/10.1016/j.schres.2019.08.026 T.T. Nguyen et al. / Schizophrenia Research xxx (xxxx) xxx 15 sequencing, shotgun metagenomics, qPCR). They all provide a varying Critical revision of the manuscript for important content and final approval of version degree of complementary, but not easily integratable information. to be published: Tanya T. Nguyen, Hugh Hathaway, Tomasz Kosciolek, Dilip V. Jeste, Rob Knight. Moreover, the microbiome is known to vary by geography. Western and Eastern populations have distinct microbiomes (Yatsunenko et al., 2012), but differences have also been shown between countries sharing Declaration of Competing Interest similar lifestyles (e.g., USA and UK) and even among different regions in The authors have declared that there are no conflicts of interest in relation to the the same country (McDonald et al., 2018). Hence, it is important to scru- subject of this study. tinize those experimental approaches in order to arrive at robust biolog- Acknowledgments ical understanding of microbial contributions to SMI. It should be This work was supported, in part, by the National Institute of Mental Health (grant emphasized, however, that despite differences across investigations, numbers K23 MH118435-01A1 to TTN, and 2R01 MH094151-06 and 5T32 MH019934- participants within each study were more homogeneous and less likely 24 to DVJ), UC San Diego Sam and Rose Stein Institute for Research on Aging, UC San Diego confounded by geographical and methodological variations, and the Center for Microbiome Innovation, University of Oxford student bursary, National Health findings of altered gut microbiome reported in each study are more Service bursary, and the Green Templeton College Learning Grant. likely driven by disease-related factors rather than geographical or tech- References nical variations. The problem occurs when trying to compare studies across different laboratories (and regions, etc.) and drawing confident Aizawa, E., Tsuji, H., Asahara, T., Takahashi, T., Teraishi, T., Yoshida, S., Ota, M., Koga, N., Hattori, K., Kunugi, H., 2016. Possible association of Bifidobacterium and Lactobacillus and generalizable conclusions about microbial contributions to SMI, in the gut microbiota of patients with major depressive disorder. J. Affect. Disord. 202, which calls for more unified methodological standards. It is understand- 254–257. able that various groups across the globe employ experimental tech- Aizawa, E., Tsuji, H., Asahara, T., Takahashi, T., Teraishi, T., Yoshida, S., Koga, N., Hattori, K., fi fi Ota, M., Kunugi, H., 2018. Bi dobacterium and Lactobacillus counts in the gut micro- niques at hand. To make ndings translatable and useful across biota of patients with bipolar disorder and healthy controls. Front. Psychiatry 9, 730. different disciplines, cohorts, and geographic regions, there is an imper- Akkasheh, G., Kashani-Poor, Z., Tajabadi-Ebrahimi, M., Jafari, P., Akbari, H., Taghizadeh, M., ative need for using consensus-based methodology for such studies. Memarzadeh, M.R., Asemi, Z., Esmaillzadeh, A., 2016. Clinical and metabolic response fi to probiotic administration in patients with major depressive disorder: a randomized, Some of the technical or geographical variation in detected speci c double-blind, placebo-controlled trial. Nutrition 32, 315–320. taxa may be mitigated through the use of functional profiles, rather Bengesser, S.A., Mörkl, S., Painold, A., Dalkner, N., Birner, A., Fellendorf, F.T., Platzer, M., than taxonomic ones (Cheung et al., 2019). Adoption of standard pro- Queissner, R., Hamm, C., Maget, A., Pilz, R., Rieger, A., Wagner-Skacel, J., cessing techniques is the only viable solution, arguably more feasible Reininghaus, B., Kapfhammer, H.P., Petek, E., Kashofer, K., Halwachs, B., Holzer, P., Waha, A., Reininghaus, E.Z., 2019. Epigenetics of the molecular clock and bacterial di- and reasonable than conducting a single worldwide multi-billion dollar versity in bipolar disorder. Psychoneuroendocrinology 101, 160–166. massive-scale study. Berk, M., Kapczinski, F., Andreazza, A.C., Dean, O.M., Giorlando, F., Maes, M., Yücel, M., An advantage of using specialized techniques, such as qPCR to quan- Gama, C.S., Dodd, S., Dean, B., Magalhães, P.V.S., Amminger, P., McGorry, P., Malhi, fi G.S., 2011. Pathways underlying neuroprogression in bipolar disorder: focus on in- tify bacteria of interest or speci c 16S rRNA primer sets, is that it helps flammation, oxidative stress and neurotrophic factors. Neurosci. Biobehav. Rev. 35, to address specific hypothesis-driven questions posed by psychiatry re- 804–817. searchers. On the other hand, the use of case-specific techniques makes Brown, S., 1997. Excess mortality of schizophrenia. A meta-analysis. Br. J. Psychiatry 171, – fi 502 508. integrating studies dif cult or impossible due to systematic variations Caporaso, J.G., Lauber, C.L., Costello, E.K., Berg-Lyons, D., Gonzalez, A., Stombaugh, J., (Debelius et al., 2016). In the microbiome field, there has been a consis- Knights, D., Gajer, P., Ravel, J., Fierer, N., Gordon, J.I., Knight, R., 2011. Moving pictures tent push for unification through the use of standardized primers and of the human microbiome. Genome Biol. 12, R50. Caporaso, J.G., Lauber, C.L., Walters, W.A., Berg-Lyons, D., Huntley, J., Fierer, N., Owens, sample preparation protocols (Caporaso et al., 2012; Knight et al., S.M., Betley, J., Fraser, L., Bauer, M., Gormley, N., Gilbert, J.A., Smith, G., Knight, R., 2018; Minich et al., 2018; Thompson et al., 2017). These universal pro- 2012. Ultra-high-throughput microbial community analysis on the Illumina HiSeq cedures result in less specificity but have been shown to provide most and MiSeq platforms. ISME J 6, 1621–1624. Carroll, I.M., Threadgill, D.W., Threadgill, D.S., 2009. The gastrointestinal microbiome: a versatility. The advantage of such an approach is the possible effect of malleable, third genome of mammals. Mamm. Genome 20, 395–403. scale that may be exemplified by the Earth Microbiome Project Casey, D.E., Hansen, T.E., Meyer, J., Nasrallah, H., 2009. Excessive mortality and morbidity (Thompson et al., 2017). This project crowd-sourced samples from the associated with schizophrenia. Medical Illness and Schizophrenia 17, 36. entire planet attempting to characterize microbial biodiversity across Castro-Nallar, E., Bendall, M.L., Pérez-Losada, M., Sabuncyan, S., Severance, E.G., Dickerson, F.B., Schroeder, J.R., Yolken, R.H., Crandall, K.A., 2015. Composition, taxonomy and all possible environments. Standardized procedures allowed for the in- functional diversity of the oropharynx microbiome in individuals with schizophrenia tegration of samples from diverse environments including soil, oceans, and controls. PeerJ 3, e1140. and human gut onto a single communal catalogue spanning over Caussy, C., Tripathi, A., Humphrey, G., Bassirian, S., Singh, S., Faulkner, C., Bettencourt, R., Rizo, E., Richards, L., Xu, Z.Z., Downes, M.R., 2019. A gut microbiome signature for cir- 20,000 samples. rhosis due to nonalcoholic fatty liver disease. Nat. Commun. 10, 1406. Another facet crucial to making substantial progress in understand- Cekanaviciute, E., Yoo, B.B., Runia, T.F., Debelius, J.W., Singh, S., Nelson, C.A., Kanner, R., ing how the gut microbiome contributes to SMI is the availability of Bencosme, Y., Lee, Y.K., Hauser, S.L., Crabtree-Hartman, E., 2017. Gut bacteria from fi multiple sclerosis patients modulate human T cells and exacerbate symptoms in data. Many researchers are justi ably hesitant to share raw data that mouse models. Proc. Natl. Acad. Sci. 114, 10713–10718. they have accumulated at great time, effort, and expense; nonetheless, Chahwan, B., Kwan, S., Isik, A., van Hemert, S., Burke, C., Roberts, L., 2019. Gut feelings: a it is necessary for reproducibility and integration efforts (Knight et al., randomised, triple-blind, placebo-controlled trial of probiotics for depressive symptoms. J. Affect. Disord. 253, 317–326. 2018). Tools such as Qiita (qiita.ucsd.edu), which is free and open- Cheung, S.G., Goldenthal, A.R., Uhlemann, A.-C., Mann, J.J., Miller, J.M., Sublette, M.E., 2019. source, make the upload, storage, processing, analysis, and sharing of Systematic review of gut microbiota and major depression. Front. Psychiatry 10, 34. data feasible (Gonzalez et al., 2018). It is a standard and requirement Chong, H.Y., Teoh, S.L., Wu, D.B.-C., Kotirum, S., Chiou, C.-F., Chaiyakunapruk, N., 2016. Global economic burden of schizophrenia: a systematic review. Neuropsychiatr. Dis. for many journals that researchers deposit data in long-term reposito- Treat. 12, 357–373. ries (e.g., NCBI-SRA, EBI-ENA, or GSA). Once those necessary steps are Christensen, K., Doblhammer, G., Rau, R., Vaupel, J.W., 2009. Ageing populations: the chal- widely adopted, using key resources like Qiita, it will it be possible to lenges ahead. Lancet 374, 1196–1208. conduct meta-analyses across different studies and arrive at robust con- Clarke, G., Grenham, S., Scully, P., Fitzgerald, P., Moloney, R.D., Shanahan, F., Dinan, T.G., Cryan, J.F., 2013. The microbiome-gut-brain axis during early life regulates the hippo- clusions providing the community with insights and hypotheses on the campal serotonergic system in a sex-dependent manner. Mol. Psychiatry 18, impact of the gut microbiome in SMI. 666–673. Coello, K., Hansen, T.H., Sørensen, N., Munkholm, K., Kessing, L.V., Pedersen, O., Vinberg, M., 2019. Gut microbiota composition in patients with newly diagnosed bipolar dis- Contributors order and their unaffected first-degree relatives. Brain Behav. Immun. 75, 112–118. Study concept and design: Tanya T. Nguyen, Tomasz Kosciolek, Dilip V. Jeste. Crumeyrolle-Arias, M., Jaglin, M., Bruneau, A., Vancassel, S., Cardona, A., Daugé, V., Naudon, L., Literature search, study screening and selection, and data extraction: Tanya T. Nguyen, Rabot, S., 2014. Absence of the gut microbiota enhances anxiety-like behavior and neu- Hugh Hathaway. roendocrine response to acute stress in rats. Psychoneuroendocrinology 42, 207–217. Data synthesis and drafting of the manuscript: Tanya T. Nguyen, Tomasz Kosciolek, Cryan, J.F., Dinan, T.G., 2012. Mind-altering microorganisms: the impact of the gut micro- Hugh Hathaway. biota on brain and behaviour. Nat. Rev. Neurosci. 13, 701–712.

Czepielewski, L., Daruy Filho, L., Brietzke, E., Grassi-Oliveira, R., 2013. Bipolar disorder and probiotic supplementation in patients with multiple sclerosis: a randomized, metabolic syndrome: a systematic review. Braz J Psychiatry 35, 88–93. double-blind, placebo-controlled trial. Clin. Nutr. 36, 1245–1249. Debelius, J., Song, S.J., Vazquez-Baeza, Y., Xu, Z.Z., Gonzalez, A., Knight, R., 2016. Tiny microbes, Lally, J., MacCabe, J.H., 2015. Antipsychotic medication in schizophrenia: a review. Br. enormous impacts: what matters in gut microbiome studies? Genome Biol. 17, 217. Med. Bull. 114, 169–179. Diaz Heijtz, R., Wang, S., Anuar, F., Qian, Y., Björkholm, B., Samuelsson, A., Hibberd, M.L., Langille, M.G.I., Zaneveld, J., Caporaso, J.G., McDonald, D., Knights, D., Reyes, J.A., Clemente, Forssberg, H., Pettersson, S., 2011. Normal gut microbiota modulates brain develop- J.C., Burkepile, D.E., Vega Thurber, R.L., Knight, R., Beiko, R.G., Huttenhower, C., 2013. ment and behavior. Proc. Natl. Acad. Sci. U. S. A. 108, 3047–3052. Predictive functional profiling of microbial communities using 16S rRNA marker gene Dickerson, F.B., Stallings, C., Origoni, A., Katsafanas, E., Savage, C.L.G., Schweinfurth, L.A.B., sequences. Nat. Biotechnol. 31, 814–821. Goga, J., Khushalani, S., Yolken, R.H., 2014. Effect of probiotic supplementation on Lee, E.E., Liu, J., Tu, X., Palmer, B.W., Eyler, L.T., Jeste, D.V., 2018. A widening longevity gap schizophrenia symptoms and association with gastrointestinal functioning: a ran- between people with schizophrenia and general population: a literature review and domized, placebo-controlled trial. Prim. Care Companion CNS Disord. 16. https:// call for action. Schizophr. Res. 196, 9–13. doi.org/10.4088/PCC.13m01579. Louis, P., Scott, K.P., Duncan, S.H., Flint, H.J., 2007. Understanding the effects of diet on bac- Dickerson, F., Adamos, M., Katsafanas, E., Khushalani, S., Origoni, A., Savage, C., terial metabolism in the large intestine. J. Appl. Microbiol. 102, 1197–1208. Schweinfurth, L., Stallings, C., Sweeney, K., Goga, J., Yolken, R.H., 2018. Adjunctive pro- Mandal, S., Van Treuren, W., White, R.A., Eggesbø, M., Knight, R., Peddada, S.D., 2015. Anal- biotic microorganisms to prevent rehospitalization in patients with acute mania: a ysis of composition of microbiomes: a novel method for studying microbial composi- randomized controlled trial. Bipolar Disord. 20, 614–621. tion. Microb. Ecol. Health Dis. 26, 27663. Dinan, T.G., Stanton, C., Cryan, J.F., 2013. Psychobiotics: a novel class of psychotropic. Biol. McDonald, D., Hyde, E., Debelius, J.W., Morton, J.T., Gonzalez, A., Ackermann, G., Aksenov, Psychiatry 74, 720–726. A.A., Behsaz, B., Brennan, C., Chen, Y., DeRight Goldasich, L., Dorrestein, P.C., Dunn, Dinan, T.G., Borre, Y.E., Cryan, J.F., 2014. Genomics of schizophrenia: time to consider the R.R., Fahimipour, A.K., Gaffney, J., Gilbert, J.A., Gogul, G., Green, J.L., Hugenholtz, P., gut microbiome? Mol. Psychiatry 19, 1252–1257. Humphrey, G., Huttenhower, C., Jackson, M.A., Janssen, S., Jeste, D.V., Jiang, L., Evans, S.J., Bassis, C.M., Hein, R., Assari, S., Flowers, S.A., Kelly, M.B., Young, V.B., Ellingrod, Kelley, S.T., Knights, D., Kosciolek, T., Ladau, J., Leach, J., Marotz, C., Meleshko, D., V.E., McInnis, M.G., 2017. The gut microbiome composition associates with bipolar Melnik, A.V., Metcalf, J.L., Mohimani, H., Montassier, E., Navas-Molina, J., Nguyen, disorder and illness severity. J. Psychiatr. Res. 87, 23–29. T.T., Peddada, S., Pevzner, P., Pollard, K.S., Rahnavard, G., Robbins-Pianka, A., Flowers, S.A., Evans, S.J., Ward, K.M., McInnis, M.G., Ellingrod, V.L., 2017. Interaction be- Sangwan, N., Shorenstein, J., Smarr, L., Song, S.J., Spector, T., Swafford, A.D., tween atypical antipsychotics and the gut microbiome in a bipolar disease cohort. Thackray, V.G., Thompson, L.R., Tripathi, A., Vázquez-Baeza, Y., Vrbanac, A., Pharmacotherapy 37, 261–267. Wischmeyer, P., Wolfe, E., Zhu, Q., American Gut Consortium, Knight, R., 2018. Flowers, S.A., Baxter, N.T., Ward, K.M., Kraal, A.Z., McInnis, M.G., Schmidt, T.M., Ellingrod, American Gut: An Open Platform for Citizen Science Microbiome Research. V.L., 2019. Effects of atypical antipsychotic treatment and resistant starch supplemen- mSystems, p. 3 https://doi.org/10.1128/mSystems.00031-18. tation on gut microbiome composition in a cohort of patients with bipolar disorder or Ménard, S., Candalh, C., Bambou, J.C., Terpend, K., Cerf-Bensussan, N., Heyman, M., 2004. schizophrenia. Pharmacotherapy 39, 161–170. Lactic acid bacteria secrete metabolites retaining anti-inflammatory properties after Gibson, G.R., Probert, H.M., Van Loo, J., Rastall, R.A., Roberfroid, M.B., 2004. Dietary modu- intestinal transport. Gut 53, 821–828. lation of the human colonic microbiota: updating the concept of prebiotics. Nutr. Res. Minich, J.J., Humphrey, G., Benitez, R.A.S., Sanders, J., Swafford, A., Allen, E.E., Knight, R., Rev. 17, 259–275. 2018. High-throughput miniaturized 16S rRNA amplicon library preparation reduces Gill, S.R., Pop, M., Deboy, R.T., Eckburg, P.B., Turnbaugh, P.J., Samuel, B.S., Gordon, J.I., costs while preserving microbiome integrity. mSystems 3. https://doi.org/10.1128/ Relman, D.A., Fraser-Liggett, C.M., Nelson, K.E., 2006. Metagenomic analysis of the mSystems.00166-18. human distal gut microbiome. Science 312, 1355–1359. Mitchell, A.J., Vancampfort, D., Sweers, K., van Winkel, R., Yu, W., De Hert, M., 2011. Prev- Gloor, G.B., Macklaim, J.M., Pawlowsky-Glahn, V., Egozcue, J.J., 2017. Microbiome datasets alence of metabolic syndrome and metabolic abnormalities in schizophrenia and re- are compositional: and this is not optional. Front. Microbiol. 8, 2224. lated disorders—a systematic review and meta-analysis. Schizophr. Bull. 39, 306–318. Gonzalez, A., Navas-Molina, J.A., Kosciolek, T., McDonald, D., Vázquez-Baeza, Y., Morton, J.T., Sanders, J., Quinn, R.A., McDonald, D., Gonzalez, A., Vázquez-Baeza, Y., Navas- Ackermann, G., DeReus, J., Janssen, S., Swafford, A.D., Orchanian, S.B., Sanders, J.G., Molina, J.A., Song, S.J., Metcalf, J.L., Hyde, E.R., Lladser, M., Dorrestein, P.C., Knight, R., Shorenstein, J., Holste, H., Petrus, S., Robbins-Pianka, A., Brislawn, C.J., Wang, M., 2017. Balance trees reveal microbial niche differentiation. mSystems 2. https://doi. Rideout, J.R., Bolyen, E., Dillon, M., Caporaso, J.G., Dorrestein, P.C., Knight, R., 2018. org/10.1128/mSystems.00162-16. Qiita: rapid, web-enabled microbiome meta-analysis. Nat. Methods 15, 796–798. Morton, J.T., Marotz, C., Washburne, A., Silverman, J., Zaramela, L.S., Edlund, A., Zengler, K., He, Y., Kosciolek, T., Tang, J., Zhou, Y., Li, Z., Ma, X., Zhu, Q., Yuan, N., Yuan, L., Li, C., Jin, K., Knight, R., 2019. Establishing microbial composition measurement standards with Knight, R., Tsuang, M.T., Chen, X., 2018. Gut microbiome and magnetic resonance reference frames. Nat. Commun. 10, 2719. spectroscopy study of subjects at ultra-high risk for psychosis may support the mem- Nagamine, T., Ido, Y., Nakamura, M., Okamura, T., 2018. 4G-β-D-galactosylsucrose as a brane hypothesis. Eur. Psychiatry 53, 37–45. prebiotics may improve underweight in inpatients with schizophrenia. Biosci Micro- Hennekens, C.H., Hennekens, A.R., Hollar, D., Casey, D.E., 2005. Schizophrenia and in- biota Food Health 37, 45–47. creased risks of cardiovascular disease. Am. Heart J. 150, 1115–1121. Naidu, A.S., Bidlack, W.R., Clemens, R.A., 1999. Probiotic spectra of lactic acid bacteria Higgins, J.A., Higbee, D.R., Donahoo, W.T., Brown, I.L., Bell, M.L., Bessesen, D.H., 2004. Re- (LAB). Crit. Rev. Food Sci. Nutr. 39, 13–126. sistant starch consumption promotes lipid oxidation. Nutr. Metab. 1, 8. Neufeld, K.M., Kang, N., Bienenstock, J., Foster, J.A., 2011. Reduced anxiety-like behavior Hill, C., Guarner, F., Reid, G., Gibson, G.R., Merenstein, D.J., Pot, B., Morelli, L., Canani, R.B., and central neurochemical change in germ-free mice: behavior in germ-free mice. Flint, H.J., Salminen, S., Calder, P.C., Sanders, M.E., 2014. The International Scientific Neurogastroenterology & Motility 23, 255–e119. Association for Probiotics and Prebiotics consensus statement on the scope and ap- Nguyen, T.T., Eyler, L.T., Jeste, D.V., 2018a. Systemic biomarkers of accelerated aging in propriate use of the term probiotic. Nat. Rev. Gastroenterol. Hepatol. 11, 506. schizophrenia: a critical review and future directions. Schizophr. Bull. 44, 398–408. Hsiao, E.Y., McBride, S.W., Hsien, S., Sharon, G., Hyde, E.R., McCue, T., Codelli, J.A., Chow, J., Nguyen, T.T., Kosciolek, T., Eyler, L.T., Knight, R., Jeste, D.V., 2018b. Overview and system- Reisman, S.E., Petrosino, J.F., Patterson, P.H., Mazmanian, S.K., 2013. Microbiota mod- atic review of studies of microbiome in schizophrenia and bipolar disorder. ulate behavioral and physiological abnormalities associated with J. Psychiatr. Res. 99, 50–61. neurodevelopmental disorders. Cell 155, 1451–1463. Nguyen, T.T., Kosciolek, T., Maldonado, Y., Daly, R.E., Martin, A.S., McDonald, D., Knight, R., Human Microbiome Project Consortium, 2012. Structure, function and diversity of the Jeste, D.V., 2019. Differences in gut microbiome composition between persons with healthy human microbiome. Nature 486, 207–214. chronic schizophrenia and healthy comparison subjects. Schizophr. Res. 204, 23–29. Jeste, D.V., Wolkowitz, O.M., Palmer, B.W., 2011. Divergent trajectories of physical, cogni- Nugent, A.P., 2005. Health properties of resistant starch. Nutr. Bull. 30, 27–54. tive, and psychosocial aging in schizophrenia. Schizophr. Bull. 37, 451–455. Okubo, R., Koga, M., Katsumata, N., Odamaki, T., Matsuyama, S., Oka, M., Narita, H., Johnston, K.L., Thomas, E.L., Bell, J.D., Frost, G.S., Robertson, M.D., 2010. Resistant starch Hashimoto, N., Kusumi, I., Xiao, J., Matsuoka, Y.J., 2019. Effect of bifidobacterium improves insulin sensitivity in metabolic syndrome. Diabet. Med. 27, 391–397. breve A-1 on anxiety and depressive symptoms in schizophrenia: a proof-of-concept Jørgensen, B.P., Krych, L., Pedersen, T.B., Plath, N., Redrobe, J.P., Hansen, A.K., Nielsen, D.S., study. J. Affect. Disord. 245, 377–385. Pedersen, C.S., Larsen, C., Sørensen, D.B., 2015. Investigating the long-term effect of Painold, A., Mörkl, S., Kashofer, K., Halwachs, B., Dalkner, N., Bengesser, S., Birner, A., subchronic phencyclidine-treatment on novel object recognition and the association Fellendorf, F., Platzer, M., Queissner, R., Schütze, G., Schwarz, M.J., Moll, N., Holzer, between the gut microbiota and behavior in the animal model of schizophrenia. P., Holl, A.K., Kapfhammer, H.-P., Gorkiewicz, G., Reininghaus, E.Z., 2019. Astep Physiol. Behav. 141, 32–39. ahead: exploring the gut microbiota in inpatients with bipolar disorder during a de- Kim, W.J., Kim, H.U., Lee, S.Y., 2017. Current state and applications of microbial genome- pressive episode. Bipolar Disord. 21, 40–49. scale metabolic models. Current Opinion in Systems Biology 2, 10–18. Palmer, B.W., Moore, R.C., Eyler, L.T., Pinto, L.L., Saks, E.R., Jeste, D.V., 2018. Avoidance of Kirkpatrick, B., Miller, B.J., 2013. Inflammation and schizophrenia. Schizophr. Bull. 39, accelerated aging in schizophrenia?: clinical and biological characterization of an ex- 1174–1179. ceptionally high functioning individual. Schizophr. Res. 196, 45–52. Kirkpatrick, B., Messias, E., Harvey, P.D., Fernandez-Egea, E., Bowie, C.R., 2008. Is schizo- Perälä, J., Suvisaari, J., Saarni, S.I., Kuoppasalmi, K., Isometsä, E., Pirkola, S., Partonen, T., phrenia a syndrome of accelerated aging? Schizophr. Bull. 34, 1024–1032. Tuulio-Henriksson, A., Hintikka, J., Kieseppä, T., Härkänen, T., Koskinen, S., Knight, R., Vrbanac, A., Taylor, B.C., Aksenov, A., Callewaert, C., Debelius, J., Gonzalez, A., Lönnqvist, J., 2007. Lifetime prevalence of psychotic and bipolar I disorders in a gen- Kosciolek, T., McCall, L.-I., McDonald, D., Melnik, A.V., Morton, J.T., Navas, J., Quinn, eral population. Arch. Gen. Psychiatry 64, 19–28. R.A., Sanders, J.G., Swafford, A.D., Thompson, L.R., Tripathi, A., Xu, Z.Z., Zaneveld, J.R., Pinto-Sanchez, M.I., Hall, G.B., Ghajar, K., Nardelli, A., Bolino, C., Lau, J.T., Martin, F.-P., Zhu, Q., Caporaso, J.G., Dorrestein, P.C., 2018. Best practices for analysing Cominetti, O., Welsh, C., Rieder, A., Traynor, J., Gregory, C., De Palma, G., Pigrau, M., microbiomes. Nat. Rev. Microbiol. 16, 410–422. Ford, A.C., Macri, J., Berger, B., Bergonzelli, G., Surette, M.G., Collins, S.M., Moayyedi, Koenig, J.E., Spor, A., Scalfone, N., Fricker, A.D., Stombaugh, J., Knight, R., Angenent, L.T., P., Bercik, P., 2017. Probiotic Bifidobacterium longum NCC3001 reduces depression Ley, R.E., 2011. Succession of microbial consortia in the developing infant gut scores and alters brain activity: a pilot study in patients with irritable bowel syn- microbiome. Proc. Natl. Acad. Sci. U. S. A. 108 (Suppl. 1), 4578–4585. drome. Gastroenterology 153, 448–459.e8. Kouchaki, E., Tamtaji, O.R., Salami, M., Bahmani, F., Daneshvar Kakhaki, R., Akbari, E., Qin, J., Li, R., Raes, J., Arumugam, M., Burgdorf, K.S., Manichanh, C., Nielsen, T., Pons, N., Tajabadi-Ebrahimi, M., Jafari, P., Asemi, Z., 2017. Clinical and metabolic response to Levenez, F., Yamada, T., Mende, D.R., Li, J., Xu, J., Li, S., Li, D., Cao, J., Wang, B., Liang,

H., Zheng, H., Xie, Y., Tap, J., Lepage, P., Bertalan, M., Batto, J.-M., Hansen, T., Le Paslier, Sudo, N., Chida, Y., Aiba, Y., Sonoda, J., Oyama, N., Yu, X.-N., Kubo, C., Koga, Y., 2004. Post- D., Linneberg, A., Nielsen, H.B., Pelletier, E., Renault, P., Sicheritz-Ponten, T., Turner, K., natal microbial colonization programs the hypothalamic—pituitary—adrenal system Zhu, H., Yu, C., Li, S., Jian, M., Zhou, Y., Li, Y., Zhang, X., Li, S., Qin, N., Yang, H., Wang, J., for stress response in mice. J. Physiol. 558, 263–275. Brunak, S., Doré, J., Guarner, F., Kristiansen, K., Pedersen, O., Parkhill, J., Weissenbach, Sun, J., Wang, F., Hong, G., Pang, M., Xu, H., Li, H., Tian, F., Fang, R., Yao, Y., Liu, J., 2016. J., MetaHIT Consortium, Bork, P., Ehrlich, S.D., Wang, J., 2010. A human gut microbial Antidepressant-like effects of sodium butyrate and its possible mechanisms of action gene catalogue established by metagenomic sequencing. Nature 464, 59–65. in mice exposed to chronic unpredictable mild stress. Neurosci. Lett. 618, 159–166. Quince, C., Walker, A.W., Simpson, J.T., Loman, N.J., Segata, N., 2017. Shotgun Teramoto, F., Rokutan, K., Kawakami, Y., Fujimura, Y., Uchida, J., Oku, K., Oka, M., metagenomics, from sampling to analysis. Nat. Biotechnol. 35, 833–844. Yoneyama, M., 1996. Effect of 4G-beta-D-galactosylsucrose (lactosucrose) on fecal Rao, A.V., Bested, A.C., Beaulne, T.M., Katzman, M.A., Iorio, C., Berardi, J.M., Logan, A.C., microflora in patients with chronic inflammatory bowel disease. J. Gastroenterol. 2009. A randomized, double-blind, placebo-controlled pilot study of a probiotic in 31, 33–39. emotional symptoms of chronic fatigue syndrome. Gut Pathog 1, 6. Thompson, L.R., Sanders, J.G., McDonald, D., Amir, A., Ladau, J., Locey, K.J., Prill, R.J., Rizzo, L.B., Costa, L.G., Mansur, R.B., Swardfager, W., Belangero, S.I., Grassi-Oliveira, R., Tripathi, A., Gibbons, S.M., Ackermann, G., Navas-Molina, J.A., Janssen, S., Kopylova, McIntyre, R.S., Bauer, M.E., Brietzke, E., 2014. The theory of bipolar disorder as an ill- E., Vázquez-Baeza, Y., González, A., Morton, J.T., Mirarab, S., Zech Xu, Z., Jiang, L., ness of accelerated aging: implications for clinical care and research. Neurosci. Haroon, M.F., Kanbar, J., Zhu, Q., Jin Song, S., Kosciolek, T., Bokulich, N.A., Lefler, J., Biobehav. Rev. 42, 157–169. Brislawn, C.J., Humphrey, G., Owens, S.M., Hampton-Marcell, J., Berg-Lyons, D., Rothschild, D., Weissbrod, O., Barkan, E., Kurilshikov, A., Korem, T., Zeevi, D., Costea, P.I., McKenzie, V., Fierer, N., Fuhrman, J.A., Clauset, A., Stevens, R.L., Shade, A., Pollard, Godneva, A., Kalka, I.N., Bar, N., Shilo, S., Lador, D., Vila, A.V., Zmora, N., Pevsner- K.S., Goodwin, K.D., Jansson, J.K., Gilbert, J.A., Knight, R., Earth Microbiome Project Fischer, M., Israeli, D., Kosower, N., Malka, G., Wolf, B.C., Avnit-Sagi, T., Lotan-Pompan, Consortium, 2017. A communal catalogue reveals Earth's multiscale microbial diver- M., Weinberger, A., Halpern, Z., Carmi, S., Fu, J., Wijmenga, C., Zhernakova, A., Elinav, sity. Nature 551, 457–463. E., Segal, E., 2018. Environment dominates over host genetics in shaping human gut Tomasik, J., Yolken, R.H., Bahn, S., Dickerson, F.B., 2015. Immunomodulatory effects of pro- microbiota. Nature 555, 210–215. biotic supplementation in schizophrenia patients: a randomized, placebo-controlled Saha, S., Chant, D., McGrath, J., 2007. A systematic review of mortality in schizophrenia: is trial. Biomark. Insights 10, 47–54. the differential mortality gap worsening over time? Arch. Gen. Psychiatry 64, Tomova, A., Husarova, V., Lakatosova, S., Bakos, J., Vlkova, B., Babinska, K., Ostatnikova, D., 1123–1131. 2015. Gastrointestinal microbiota in children with autism in Slovakia. Physiol. Behav. Sampson, T.R., Debelius, J.W., Thron, T., Janssen, S., Shastri, G.G., Ilhan, Z.E., Challis, C., 138, 179–187. Schretter, C.E., Rocha, S., Gradinaru, V., Chesselet, M.-F., Keshavarzian, A., Shannon, Turnbaugh, P.J., Ley, R.E., Hamady, M., Fraser-Liggett, C.M., Knight, R., Gordon, J.I., 2007. K.M., Krajmalnik-Brown, R., Wittung-Stafshede, P., Knight, R., Mazmanian, S.K., The human microbiome project. Nature 449, 804–810. 2016. Gut microbiota regulate motor deficits and neuroinflammation in a model of Ursell, L.K., Haiser, H.J., Van Treuren, W., Garg, N., Reddivari, L., Vanamala, J., Dorrestein, Parkinson's disease. Cell 167, 1469–1480.e12. P.C., Turnbaugh, P.J., Knight, R., 2014. The intestinal metabolome: an intersection be- Sarkar, A., Lehto, S.M., Harty, S., Dinan, T.G., Cryan, J.F., Burnet, P.W.J., 2016. Psychobiotics tween microbiota and host. Gastroenterology 146, 1470–1476. and the manipulation of bacteria–gut–brain signals. Trends Neurosci. 39, 763–781. Vinberg, M., Ottesen, N.M., Meluken, I., Sørensen, N., Pedersen, O., Kessing, L.V., Savignac, H.M., Corona, G., Mills, H., Chen, L., Spencer, J.P.E., Tzortzis, G., Burnet, P.W.J., Miskowiak, K.W., 2019. Remitted affective disorders and high familial risk of affective 2013. Prebiotic feeding elevates central brain derived neurotrophic factor, N- disorders associate with aberrant intestinal microbiota. Acta Psychiatr. Scand. 139, methyl-D-aspartate receptor subunits and D-serine. Neurochem. Int. 63, 756–764. 174–184. Schiffrin, E.J., Brassart, D., Servin, A.L., Rochat, F., Donnet-Hughes, A., 1997. Immune mod- Viron, M.J., Stern, T.A., 2010. The impact of serious mental illness on health and ulation of blood leukocytes in humans by lactic acid bacteria: criteria for strain selec- healthcare. Psychosomatics 51, 458–465. tion. Am. J. Clin. Nutr. 66, 515S–520S. Walker, E.R., McGee, R.E., Druss, B.G., 2015. Mortality in mental disorders and global dis- Schmidt, T.S., Hayward, M.R., Coelho, L.P., Li, S.S., Costea, P.I., Voigt, A.Y., Wirbel, J., ease burden implications: a systematic review and meta-analysis. JAMA Psychiatry Maistrenko, O.M., Alves, R.J., Bergsten, E., de Beaufort, C., Sobhani, I., Heintz- 72, 334–341. Buschart, A., Sunagawa, S., Zeller, G., Wilmes, P., Bork, P., 2019. Extensive transmis- Wall, R., Cryan, J.F., Ross, R.P., Fitzgerald, G.F., Dinan, T.G., Stanton, C., 2014. Bacterial neu- sion of microbes along the gastrointestinal tract. Elife 8. https://doi.org/10.7554/ roactive compounds produced by psychobiotics. Adv. Exp. Med. Biol. 817, 221–239. eLife.42693. Weiss, S., Xu, Z.Z., Peddada, S., Amir, A., Bittinger, K., Gonzalez, A., Lozupone, C., Zaneveld, Schwarz, E., Maukonen, J., Hyytiäinen, T., Kieseppä, T., Orešič, M., Sabunciyan, S., Mantere, J.R., Vázquez-Baeza, Y., Birmingham, A., Hyde, E.R., Knight, R., 2017. Normalization O., Saarela, M., Yolken, R., Suvisaari, J., 2018. Analysis of microbiota in first episode and microbial differential abundance strategies depend upon data characteristics. psychosis identifies preliminary associations with symptom severity and treatment Microbiome 5, 27. response. Schizophr. Res. 192, 398–403. Whiteford, H.A., Degenhardt, L., Rehm, J., Baxter, A.J., Ferrari, A.J., Erskine, H.E., Charlson, Severance, E.G., Alaedini, A., Yang, S., Halling, M., Gressitt, K.L., Stallings, C.R., Origoni, A.E., F.J., Norman, R.E., Flaxman, A.D., Johns, N., Burstein, R., Murray, C.J.L., Vos, T., 2013. Vaughan, C., Khushalani, S., Leweke, F.M., Dickerson, F.B., Yolken, R.H., 2012. Gastro- Global burden of disease attributable to mental and substance use disorders: findings intestinal inflammation and associated immune activation in schizophrenia. from the Global Burden of Disease Study 2010. Lancet 382, 1575–1586. Schizophr. Res. 138, 48–53. Yatsunenko, T., Rey, F.E., Manary, M.J., Trehan, I., Dominguez-Bello, M.G., Contreras, M., Severance, E.G., Gressitt, K.L., Stallings, C.R., Origoni, A.E., Khushalani, S., Leweke, F.M., Magris, M., Hidalgo, G., Baldassano, R.N., Anokhin, A.P., Heath, A.C., Warner, B., Dickerson, F.B., Yolken, R.H., 2013. Discordant patterns of bacterial translocation Reeder, J., Kuczynski, J., Caporaso, J.G., Lozupone, C.A., Lauber, C., Clemente, J.C., markers and implications for innate immune imbalances in schizophrenia. Schizophr. Knights, D., Knight, R., Gordon, J.I., 2012. Human gut microbiome viewed across age Res. 148, 130–137. and geography. Nature 486, 222–227. Severance, E.G., Gressitt, K.L., Yang, S., Stallings, C.R., Origoni, A.E., Vaughan, C., Khushalani, Yolken, R.H., Severance, E.G., Sabunciyan, S., Gressitt, K.L., Chen, O., Stallings, C., Origoni, A., S., Alaedini, A., Dickerson, F.B., Yolken, R.H., 2014. Seroreactive marker for inflamma- Katsafanas, E., Schweinfurth, L.A.B., Savage, C.L.G., Banis, M., Khushalani, S., Dickerson, tory bowel disease and associations with antibodies to dietary proteins in bipolar dis- F.B., 2015. Metagenomic sequencing indicates that the oropharyngeal phageome of order. Bipolar Disord. 16, 230–240. individuals with schizophrenia differs from that of controls. Schizophr. Bull. 41, Severance, E.G., Gressitt, K.L., Stallings, C.R., Katsafanas, E., Schweinfurth, L.A., Savage, 1153–1161. C.L.G., Adamos, M.B., Sweeney, K.M., Origoni, A.E., Khushalani, S., Dickerson, F.B., Yuan, X., Zhang, P., Wang, Y., Liu, Y., Li, X., Kumar, B.U., Hei, G., Lv, L., Huang, X.-F., Fan, X., Yolken, R.H., 2017. Probiotic normalization of Candida albicans in schizophrenia: a Song, X., 2018. Changes in metabolism and microbiota after 24-week risperidone randomized, placebo-controlled, longitudinal pilot study. Brain Behav. Immun. 62, treatment in drug naïve, normal weight patients with first episode schizophrenia. 41–45. Schizophr. Res. 201, 299–306. Sheih, Y.H., Chiang, B.L., Wang, L.H., Liao, C.K., Gill, H.S., 2001. Systemic immunity- Zheng, P., Zeng, B., Zhou, C., Liu, M., Fang, Z., Xu, X., Zeng, L., Chen, J., Fan, S., Du, X., Zhang, enhancing effects in healthy subjects following dietary consumption of the lactic X., Yang, D., Yang, Y., Meng, H., Li, W., Melgiri, N.D., Licinio, J., Wei, H., Xie, P., 2016. Gut acid bacterium Lactobacillus rhamnosus HN001. J. Am. Coll. Nutr. 20, 149–156. microbiome remodeling induces depressive-like behaviors through a pathway medi- Shen, Y., Xu, J., Li, Z., Huang, Y., Yuan, Y., Wang, J., Zhang, M., Hu, S., Liang, Y., 2018. Anal- ated by the host's metabolism. Mol. Psychiatry 21, 786–796. ysis of gut microbiota diversity and auxiliary diagnosis as a biomarker in patients Zheng, P., Zeng, B., Liu, M., Chen, J., Pan, J., Han, Y., Liu, Y., Cheng, K., Zhou, C., Wang, H., with schizophrenia: a cross-sectional study. Schizophr. Res. https://doi.org/10.1016/ Zhou, X., Gui, S., Perry, S.W., Wong, M.-L., Licinio, J., Wei, H., Xie, P., 2019. The gut j.schres.2018.01.002. microbiome from patients with schizophrenia modulates the glutamate-glutamine- Soreca, I., Fagiolini, A., Frank, E., Houck, P.R., Thompson, W.K., Kupfer, D.J., 2008. Relation- GABA cycle and schizophrenia-relevant behaviors in mice. Sci. Adv. 5, eaau8317. ship of general medical burden, duration of illness and age in patients with bipolar I Zhou, L., Foster, J.A., 2015. Psychobiotics and the gut-brain axis: in the pursuit of happi- disorder. J. Psychiatr. Res. 42, 956–961. ness. Neuropsychiatr. Dis. Treat. 11, 715–723.